diff --git a/ppcls/arch/backbone/__init__.py b/ppcls/arch/backbone/__init__.py index d7159fd80..f64a3ae58 100644 --- a/ppcls/arch/backbone/__init__.py +++ b/ppcls/arch/backbone/__init__.py @@ -29,7 +29,7 @@ from .legendary_models.pp_hgnet import PPHGNet_tiny, PPHGNet_small, PPHGNet_base from .model_zoo.resnet_vc import ResNet50_vc from .model_zoo.resnext import ResNeXt50_32x4d, ResNeXt50_64x4d, ResNeXt101_32x4d, ResNeXt101_64x4d, ResNeXt152_32x4d, ResNeXt152_64x4d from .model_zoo.resnext_vd import ResNeXt50_vd_32x4d, ResNeXt50_vd_64x4d, ResNeXt101_vd_32x4d, ResNeXt101_vd_64x4d, ResNeXt152_vd_32x4d, ResNeXt152_vd_64x4d -from .model_zoo.res2net import Res2Net50_26w_4s, Res2Net50_14w_8s +from .model_zoo.res2net import Res2Net50_26w_4s, Res2Net50_14w_8s, Res2Net50_26w_6s, Res2Net50_26w_8s, Res2Net50_48w_2s, Res2Net101_26w_4s from .model_zoo.res2net_vd import Res2Net50_vd_26w_4s, Res2Net101_vd_26w_4s, Res2Net200_vd_26w_4s from .model_zoo.se_resnet_vd import SE_ResNet18_vd, SE_ResNet34_vd, SE_ResNet50_vd from .model_zoo.se_resnext_vd import SE_ResNeXt50_vd_32x4d, SE_ResNeXt50_vd_32x4d, SENet154_vd @@ -37,7 +37,7 @@ from .model_zoo.se_resnext import SE_ResNeXt50_32x4d, SE_ResNeXt101_32x4d, SE_Re from .model_zoo.dpn import DPN68, DPN92, DPN98, DPN107, DPN131 from .model_zoo.densenet import DenseNet121, DenseNet161, DenseNet169, DenseNet201, DenseNet264 from .model_zoo.efficientnet import EfficientNetB0, EfficientNetB1, EfficientNetB2, EfficientNetB3, EfficientNetB4, EfficientNetB5, EfficientNetB6, EfficientNetB7, EfficientNetB0_small -from .model_zoo.resnest import ResNeSt50_fast_1s1x64d, ResNeSt50, ResNeSt101 +from .model_zoo.resnest import ResNeSt50_fast_1s1x64d, ResNeSt50, ResNeSt101, ResNeSt200, ResNeSt269 from .model_zoo.googlenet import GoogLeNet from .model_zoo.mobilenet_v2 import MobileNetV2_x0_25, MobileNetV2_x0_5, MobileNetV2_x0_75, MobileNetV2, MobileNetV2_x1_5, MobileNetV2_x2_0 from .model_zoo.shufflenet_v2 import ShuffleNetV2_x0_25, ShuffleNetV2_x0_33, ShuffleNetV2_x0_5, ShuffleNetV2_x1_0, ShuffleNetV2_x1_5, ShuffleNetV2_x2_0, ShuffleNetV2_swish @@ -49,7 +49,7 @@ from .model_zoo.xception_deeplab import Xception41_deeplab, Xception65_deeplab from .model_zoo.resnext101_wsl import ResNeXt101_32x8d_wsl, ResNeXt101_32x16d_wsl, ResNeXt101_32x32d_wsl, ResNeXt101_32x48d_wsl from .model_zoo.squeezenet import SqueezeNet1_0, SqueezeNet1_1 from .model_zoo.darknet import DarkNet53 -from .model_zoo.regnet import RegNetX_200MF, RegNetX_4GF, RegNetX_32GF, RegNetY_200MF, RegNetY_4GF, RegNetY_32GF +from .model_zoo.regnet import RegNetX_200MF, RegNetX_400MF, RegNetX_600MF, RegNetX_800MF, RegNetX_1600MF, RegNetX_3200MF, RegNetX_4GF, RegNetX_6400MF, RegNetX_8GF, RegNetX_12GF, RegNetX_16GF, RegNetX_32GF from .model_zoo.vision_transformer import ViT_small_patch16_224, ViT_base_patch16_224, ViT_base_patch16_384, ViT_base_patch32_384, ViT_large_patch16_224, ViT_large_patch16_384, ViT_large_patch32_384 from .model_zoo.distilled_vision_transformer import DeiT_tiny_patch16_224, DeiT_small_patch16_224, DeiT_base_patch16_224, DeiT_tiny_distilled_patch16_224, DeiT_small_distilled_patch16_224, DeiT_base_distilled_patch16_224, DeiT_base_patch16_384, DeiT_base_distilled_patch16_384 from .legendary_models.swin_transformer import SwinTransformer_tiny_patch4_window7_224, SwinTransformer_small_patch4_window7_224, SwinTransformer_base_patch4_window7_224, SwinTransformer_base_patch4_window12_384, SwinTransformer_large_patch4_window7_224, SwinTransformer_large_patch4_window12_384 @@ -60,15 +60,15 @@ from .model_zoo.gvt import pcpvt_small, pcpvt_base, pcpvt_large, alt_gvt_small, from .model_zoo.levit import LeViT_128S, LeViT_128, LeViT_192, LeViT_256, LeViT_384 from .model_zoo.dla import DLA34, DLA46_c, DLA46x_c, DLA60, DLA60x, DLA60x_c, DLA102, DLA102x, DLA102x2, DLA169 from .model_zoo.rednet import RedNet26, RedNet38, RedNet50, RedNet101, RedNet152 -from .model_zoo.tnt import TNT_small +from .model_zoo.tnt import TNT_small, TNT_base from .model_zoo.hardnet import HarDNet68, HarDNet85, HarDNet39_ds, HarDNet68_ds from .model_zoo.cspnet import CSPDarkNet53 from .model_zoo.pvt_v2 import PVT_V2_B0, PVT_V2_B1, PVT_V2_B2_Linear, PVT_V2_B2, PVT_V2_B3, PVT_V2_B4, PVT_V2_B5 from .model_zoo.mobilevit import MobileViT_XXS, MobileViT_XS, MobileViT_S -from .model_zoo.repvgg import RepVGG_A0, RepVGG_A1, RepVGG_A2, RepVGG_B0, RepVGG_B1, RepVGG_B2, RepVGG_B1g2, RepVGG_B1g4, RepVGG_B2g4, RepVGG_B3g4 -from .model_zoo.van import VAN_B0 +from .model_zoo.repvgg import RepVGG_A0, RepVGG_A1, RepVGG_A2, RepVGG_B0, RepVGG_B1, RepVGG_B2, RepVGG_B1g2, RepVGG_B1g4, RepVGG_B2g4, RepVGG_B3, RepVGG_B3g4, RepVGG_D2se +from .model_zoo.van import VAN_B0, VAN_B1, VAN_B2, VAN_B3 from .model_zoo.peleenet import PeleeNet -from .model_zoo.convnext import ConvNeXt_tiny +from .model_zoo.convnext import ConvNeXt_tiny, ConvNeXt_small, ConvNeXt_base_224, ConvNeXt_base_384, ConvNeXt_large_224, ConvNeXt_large_384 from .model_zoo.nextvit import NextViT_small_224, NextViT_base_224, NextViT_large_224, NextViT_small_384, NextViT_base_384, NextViT_large_384 from .model_zoo.cae import cae_base_patch16_224, cae_large_patch16_224 diff --git a/ppcls/arch/backbone/model_zoo/regnet.py b/ppcls/arch/backbone/model_zoo/regnet.py index 151e39439..45add57ce 100644 --- a/ppcls/arch/backbone/model_zoo/regnet.py +++ b/ppcls/arch/backbone/model_zoo/regnet.py @@ -56,12 +56,6 @@ MODEL_URLS = { "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RegNetX_16GF_pretrained.pdparams", "RegNetX_32GF": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RegNetX_32GF_pretrained.pdparams", - "RegNetY_200MF": - "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RegNetY_200MF_pretrained.pdparams", - "RegNetY_4GF": - "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RegNetY_4GF_pretrained.pdparams", - "RegNetY_32GF": - "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RegNetY_32GF_pretrained.pdparams" } __all__ = list(MODEL_URLS.keys()) diff --git a/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_224.yaml b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_224.yaml new file mode 100644 index 000000000..a0630335b --- /dev/null +++ b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_224.yaml @@ -0,0 +1,170 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + update_freq: 4 # for 8 cards + +# model ema +EMA: + decay: 0.9999 + + +# model architecture +Arch: + name: ConvNeXt_base_224 + class_num: 1000 + drop_path_rate: 0.1 + layer_scale_init_value: 1e-6 + head_init_scale: 1.0 + + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + # for 8 cards + name: Cosine + learning_rate: 4e-3 # lr 4e-3 for total_batch_size 4096 + eta_min: 1e-6 + warmup_epoch: 20 + warmup_start_lr: 0 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: bicubic + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: bicubic + img_size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_384.yaml b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_384.yaml new file mode 100644 index 000000000..bbe9a9a89 --- /dev/null +++ b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_base_384.yaml @@ -0,0 +1,170 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 384, 384] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + update_freq: 4 # for 8 cards + +# model ema +EMA: + decay: 0.9999 + + +# model architecture +Arch: + name: ConvNeXt_base_384 + class_num: 1000 + drop_path_rate: 0.1 + layer_scale_init_value: 1e-6 + head_init_scale: 1.0 + + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + # for 8 cards + name: Cosine + learning_rate: 4e-3 # lr 4e-3 for total_batch_size 4096 + eta_min: 1e-6 + warmup_epoch: 20 + warmup_start_lr: 0 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 384 + interpolation: bicubic + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: bicubic + img_size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 384 + interpolation: bicubic + backend: pil + - CropImage: + size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 384 + interpolation: bicubic + backend: pil + - CropImage: + size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_224.yaml b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_224.yaml new file mode 100644 index 000000000..e15cbf6f6 --- /dev/null +++ b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_224.yaml @@ -0,0 +1,170 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + update_freq: 4 # for 8 cards + +# model ema +EMA: + decay: 0.9999 + + +# model architecture +Arch: + name: ConvNeXt_large_224 + class_num: 1000 + drop_path_rate: 0.1 + layer_scale_init_value: 1e-6 + head_init_scale: 1.0 + + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + # for 8 cards + name: Cosine + learning_rate: 4e-3 # lr 4e-3 for total_batch_size 4096 + eta_min: 1e-6 + warmup_epoch: 20 + warmup_start_lr: 0 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: bicubic + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: bicubic + img_size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_384.yaml b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_384.yaml new file mode 100644 index 000000000..41e669885 --- /dev/null +++ b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_large_384.yaml @@ -0,0 +1,170 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 384, 384] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + update_freq: 4 # for 8 cards + +# model ema +EMA: + decay: 0.9999 + + +# model architecture +Arch: + name: ConvNeXt_large_384 + class_num: 1000 + drop_path_rate: 0.1 + layer_scale_init_value: 1e-6 + head_init_scale: 1.0 + + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + # for 8 cards + name: Cosine + learning_rate: 4e-3 # lr 4e-3 for total_batch_size 4096 + eta_min: 1e-6 + warmup_epoch: 20 + warmup_start_lr: 0 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 384 + interpolation: bicubic + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: bicubic + img_size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 384 + interpolation: bicubic + backend: pil + - CropImage: + size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 384 + interpolation: bicubic + backend: pil + - CropImage: + size: 384 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_small.yaml b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_small.yaml new file mode 100644 index 000000000..b929023d5 --- /dev/null +++ b/ppcls/configs/ImageNet/ConvNeXt/ConvNeXt_small.yaml @@ -0,0 +1,170 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + update_freq: 4 # for 8 cards + +# model ema +EMA: + decay: 0.9999 + + +# model architecture +Arch: + name: ConvNeXt_small + class_num: 1000 + drop_path_rate: 0.1 + layer_scale_init_value: 1e-6 + head_init_scale: 1.0 + + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + # for 8 cards + name: Cosine + learning_rate: 4e-3 # lr 4e-3 for total_batch_size 4096 + eta_min: 1e-6 + warmup_epoch: 20 + warmup_start_lr: 0 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: bicubic + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: bicubic + img_size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_12GF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_12GF.yaml new file mode 100644 index 000000000..7bd4b2f3f --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_12GF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_12GF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_1600MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_1600MF.yaml new file mode 100644 index 000000000..476934f92 --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_1600MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_1600MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_16GF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_16GF.yaml new file mode 100644 index 000000000..36e2d4edf --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_16GF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_16GF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_200MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_200MF.yaml new file mode 100644 index 000000000..83c0a3a47 --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_200MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_200MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_3200MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_3200MF.yaml new file mode 100644 index 000000000..cacedfbc2 --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_3200MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetY_3200MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_32GF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_32GF.yaml new file mode 100644 index 000000000..412f1d73d --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_32GF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_32GF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_400MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_400MF.yaml new file mode 100644 index 000000000..148ee1e2c --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_400MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_400MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_600MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_600MF.yaml new file mode 100644 index 000000000..948c133cd --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_600MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_600MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_6400MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_6400MF.yaml new file mode 100644 index 000000000..fd1a38d4d --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_6400MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_6400MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_800MF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_800MF.yaml new file mode 100644 index 000000000..dfd6a4c7b --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_800MF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_800MF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RegNet/RegNetX_8GF.yaml b/ppcls/configs/ImageNet/RegNet/RegNetX_8GF.yaml new file mode 100644 index 000000000..6353479ce --- /dev/null +++ b/ppcls/configs/ImageNet/RegNet/RegNetX_8GF.yaml @@ -0,0 +1,130 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RegNetX_8GF + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] diff --git a/ppcls/configs/ImageNet/RepVGG/RepVGG_B3.yaml b/ppcls/configs/ImageNet/RepVGG/RepVGG_B3.yaml new file mode 100644 index 000000000..9933f0554 --- /dev/null +++ b/ppcls/configs/ImageNet/RepVGG/RepVGG_B3.yaml @@ -0,0 +1,137 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 200 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RepVGG_B3 + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Cosine + learning_rate: 0.1 + warmup_epoch: 5 + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - AutoAugment: + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + batch_transform_ops: + - MixupOperator: + alpha: 0.2 + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 256 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/RepVGG/RepVGG_B3g4.yaml b/ppcls/configs/ImageNet/RepVGG/RepVGG_B3g4.yaml index 15005149e..b5f3780da 100644 --- a/ppcls/configs/ImageNet/RepVGG/RepVGG_B3g4.yaml +++ b/ppcls/configs/ImageNet/RepVGG/RepVGG_B3g4.yaml @@ -86,6 +86,8 @@ DataLoader: channel_first: False - ResizeImage: resize_short: 256 + interpolation: bicubic + backend: pil - CropImage: size: 224 - NormalizeImage: @@ -95,7 +97,7 @@ DataLoader: order: '' sampler: name: DistributedBatchSampler - batch_size: 64 + batch_size: 128 drop_last: False shuffle: False loader: @@ -111,6 +113,8 @@ Infer: channel_first: False - ResizeImage: resize_short: 256 + interpolation: bicubic + backend: pil - CropImage: size: 224 - NormalizeImage: @@ -130,4 +134,4 @@ Metric: topk: [1, 5] Eval: - TopkAcc: - topk: [1, 5] + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/RepVGG/RepVGG_D2se.yaml b/ppcls/configs/ImageNet/RepVGG/RepVGG_D2se.yaml new file mode 100644 index 000000000..715f59916 --- /dev/null +++ b/ppcls/configs/ImageNet/RepVGG/RepVGG_D2se.yaml @@ -0,0 +1,137 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 200 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 320, 320] + save_inference_dir: ./inference + +# model architecture +Arch: + name: RepVGG_D2se + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Cosine + learning_rate: 0.1 + warmup_epoch: 5 + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - AutoAugment: + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + batch_transform_ops: + - MixupOperator: + alpha: 0.2 + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 320 + interpolation: bicubic + backend: pil + - CropImage: + size: 320 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 320 + interpolation: bicubic + backend: pil + - CropImage: + size: 320 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ResNeSt/ResNeSt101.yaml b/ppcls/configs/ImageNet/ResNeSt/ResNeSt101.yaml index d99e885e3..188c3e62a 100644 --- a/ppcls/configs/ImageNet/ResNeSt/ResNeSt101.yaml +++ b/ppcls/configs/ImageNet/ResNeSt/ResNeSt101.yaml @@ -38,7 +38,7 @@ Optimizer: learning_rate: 0.1 regularizer: name: 'L2' - coeff: 0.00007 + coeff: 0.0001 # data loader for train and eval @@ -95,7 +95,7 @@ DataLoader: order: '' sampler: name: DistributedBatchSampler - batch_size: 64 + batch_size: 128 drop_last: False shuffle: False loader: @@ -128,4 +128,4 @@ Metric: Train: Eval: - TopkAcc: - topk: [1, 5] + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ResNeSt/ResNeSt200.yaml b/ppcls/configs/ImageNet/ResNeSt/ResNeSt200.yaml new file mode 100644 index 000000000..83c454814 --- /dev/null +++ b/ppcls/configs/ImageNet/ResNeSt/ResNeSt200.yaml @@ -0,0 +1,131 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: ResNeSt200 + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Cosine + learning_rate: 0.1 + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 256 + - RandFlipImage: + flip_code: 1 + - AutoAugment: + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + batch_transform_ops: + - MixupOperator: + alpha: 0.2 + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 288 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 288 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/ResNeSt/ResNeSt269.yaml b/ppcls/configs/ImageNet/ResNeSt/ResNeSt269.yaml new file mode 100644 index 000000000..3cc0cb16e --- /dev/null +++ b/ppcls/configs/ImageNet/ResNeSt/ResNeSt269.yaml @@ -0,0 +1,131 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 256, 256] + save_inference_dir: ./inference + +# model architecture +Arch: + name: ResNeSt269 + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Cosine + learning_rate: 0.1 + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 256 + - RandFlipImage: + flip_code: 1 + - AutoAugment: + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + batch_transform_ops: + - MixupOperator: + alpha: 0.2 + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 288 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 288 + - CropImage: + size: 256 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/TNT/TNT_base.yaml b/ppcls/configs/ImageNet/TNT/TNT_base.yaml new file mode 100644 index 000000000..a93c07e76 --- /dev/null +++ b/ppcls/configs/ImageNet/TNT/TNT_base.yaml @@ -0,0 +1,134 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 120 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + +# model architecture +Arch: + name: TNT_base + class_num: 1000 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + Eval: + - CELoss: + weight: 1.0 + + +Optimizer: + name: Momentum + momentum: 0.9 + lr: + name: Piecewise + learning_rate: 0.1 + decay_epochs: [30, 60, 90] + values: [0.1, 0.01, 0.001, 0.0001] + regularizer: + name: 'L2' + coeff: 0.0001 + + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + - RandFlipImage: + flip_code: 1 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + + sampler: + name: DistributedBatchSampler + batch_size: 64 + drop_last: False + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Train: + - TopkAcc: + topk: [1, 5] + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/TNT/TNT_small.yaml b/ppcls/configs/ImageNet/TNT/TNT_small.yaml index 1eab42306..7a6ec687b 100644 --- a/ppcls/configs/ImageNet/TNT/TNT_small.yaml +++ b/ppcls/configs/ImageNet/TNT/TNT_small.yaml @@ -83,6 +83,8 @@ DataLoader: channel_first: False - ResizeImage: resize_short: 248 + interpolation: bicubic + backend: pil - CropImage: size: 224 - NormalizeImage: @@ -92,7 +94,7 @@ DataLoader: order: '' sampler: name: DistributedBatchSampler - batch_size: 64 + batch_size: 128 drop_last: False shuffle: False loader: @@ -108,6 +110,8 @@ Infer: channel_first: False - ResizeImage: resize_short: 248 + interpolation: bicubic + backend: pil - CropImage: size: 224 - NormalizeImage: @@ -127,4 +131,4 @@ Metric: topk: [1, 5] Eval: - TopkAcc: - topk: [1, 5] + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/VAN/VAN_B1.yaml b/ppcls/configs/ImageNet/VAN/VAN_B1.yaml new file mode 100644 index 000000000..a31f33fcf --- /dev/null +++ b/ppcls/configs/ImageNet/VAN/VAN_B1.yaml @@ -0,0 +1,158 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + +# model architecture +Arch: + name: VAN_B1 + class_num: 1000 + drop_path_rate: 0.1 + drop_rate: 0.0 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + name: Cosine + learning_rate: 1e-3 + eta_min: 1e-6 + warmup_epoch: 5 + warmup_start_lr: 1e-6 + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: random + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: random + img_size: 224 + mean: [0.5, 0.5, 0.5] + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 256 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/VAN/VAN_B2.yaml b/ppcls/configs/ImageNet/VAN/VAN_B2.yaml new file mode 100644 index 000000000..21ff3716e --- /dev/null +++ b/ppcls/configs/ImageNet/VAN/VAN_B2.yaml @@ -0,0 +1,158 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + +# model architecture +Arch: + name: VAN_B2 + class_num: 1000 + drop_path_rate: 0.1 + drop_rate: 0.0 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + name: Cosine + learning_rate: 1e-3 + eta_min: 1e-6 + warmup_epoch: 5 + warmup_start_lr: 1e-6 + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: random + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: random + img_size: 224 + mean: [0.5, 0.5, 0.5] + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 256 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file diff --git a/ppcls/configs/ImageNet/VAN/VAN_B3.yaml b/ppcls/configs/ImageNet/VAN/VAN_B3.yaml new file mode 100644 index 000000000..6db5fbbac --- /dev/null +++ b/ppcls/configs/ImageNet/VAN/VAN_B3.yaml @@ -0,0 +1,158 @@ +# global configs +Global: + checkpoints: null + pretrained_model: null + output_dir: ./output/ + device: gpu + save_interval: 1 + eval_during_train: True + eval_interval: 1 + epochs: 300 + print_batch_step: 10 + use_visualdl: False + # used for static mode and model export + image_shape: [3, 224, 224] + save_inference_dir: ./inference + # training model under @to_static + to_static: False + +# model architecture +Arch: + name: VAN_B3 + class_num: 1000 + drop_path_rate: 0.1 + drop_rate: 0.0 + +# loss function config for traing/eval process +Loss: + Train: + - CELoss: + weight: 1.0 + epsilon: 0.1 + Eval: + - CELoss: + weight: 1.0 + +Optimizer: + name: AdamW + beta1: 0.9 + beta2: 0.999 + epsilon: 1e-8 + weight_decay: 0.05 + one_dim_param_no_weight_decay: True + lr: + name: Cosine + learning_rate: 1e-3 + eta_min: 1e-6 + warmup_epoch: 5 + warmup_start_lr: 1e-6 + +# data loader for train and eval +DataLoader: + Train: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/train_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - RandCropImage: + size: 224 + interpolation: random + backend: pil + - RandFlipImage: + flip_code: 1 + - TimmAutoAugment: + config_str: rand-m9-mstd0.5-inc1 + interpolation: random + img_size: 224 + mean: [0.5, 0.5, 0.5] + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - RandomErasing: + EPSILON: 0.25 + sl: 0.02 + sh: 1.0/3.0 + r1: 0.3 + attempt: 10 + use_log_aspect: True + mode: pixel + batch_transform_ops: + - OpSampler: + MixupOperator: + alpha: 0.8 + prob: 0.5 + CutmixOperator: + alpha: 1.0 + prob: 0.5 + sampler: + name: DistributedBatchSampler + batch_size: 256 + drop_last: True + shuffle: True + loader: + num_workers: 4 + use_shared_memory: True + + Eval: + dataset: + name: ImageNetDataset + image_root: ./dataset/ILSVRC2012/ + cls_label_path: ./dataset/ILSVRC2012/val_list.txt + transform_ops: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + sampler: + name: DistributedBatchSampler + batch_size: 128 + drop_last: False + shuffle: False + loader: + num_workers: 4 + use_shared_memory: True + +Infer: + infer_imgs: docs/images/inference_deployment/whl_demo.jpg + batch_size: 10 + transforms: + - DecodeImage: + to_rgb: True + channel_first: False + - ResizeImage: + resize_short: 248 + interpolation: bicubic + backend: pil + - CropImage: + size: 224 + - NormalizeImage: + scale: 1.0/255.0 + mean: [0.5, 0.5, 0.5] + std: [0.5, 0.5, 0.5] + order: '' + - ToCHWImage: + PostProcess: + name: Topk + topk: 5 + class_id_map_file: ppcls/utils/imagenet1k_label_list.txt + +Metric: + Eval: + - TopkAcc: + topk: [1, 5] \ No newline at end of file