bugfix_swintiny (#79)

1. add easycv_model default_path
2. bugfix pretrained

configs/classification/cifar10/swintiny_b64_5e_jpg.py

@@ -6,7 +6,6 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='swin_tiny_patch4_window7_224',
         num_classes=10,
-        pretrained=True,
     ),
     head=dict(type='ClsHead', with_fc=False))
 # dataset settings

configs/classification/imagenet/swint/imagenet_swin_base_patch4_window7_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='swin_base_patch4_window7_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/swint/imagenet_swin_large_patch4_window7_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='swin_large_patch4_window7_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/swint/imagenet_swin_tiny_patch4_window7_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='swin_tiny_patch4_window7_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/swint/swin_tiny_patch4_window7_224_b64x16_300e_jpg.py

@@ -9,6 +9,7 @@ log_config = dict(
 model = dict(
     type='Classification',
     train_preprocess=['mixUp'],
+    pretrained=False,
     mixup_cfg=dict(
         mixup_alpha=0.8,
         cutmix_alpha=1.0,
@@ -20,7 +21,6 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='swin_tiny_patch4_window7_224',
         num_classes=1000,
-        pretrained=False,
     ),
     head=dict(
         type='ClsHead',

configs/classification/imagenet/timm/timm_config.py

@@ -9,7 +9,6 @@ log_config = dict(
 model = dict(
     type='Classification',
     train_preprocess=['mixUp'],
-    pretrained=True,
     mixup_cfg=dict(
         mixup_alpha=0.2,
         prob=1.0,

configs/classification/imagenet/vit/imagenet_vit_base_patch16_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='vit_base_patch16_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/vit/imagenet_vit_base_patch32_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='vit_base_patch32_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/vit/imagenet_vit_large_patch16_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='vit_large_patch16_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/vit/imagenet_vit_large_patch32_224_jpg.py

@@ -5,5 +5,4 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='vit_large_patch32_224',
         num_classes=1000,
-        pretrained=False,
     ))

configs/classification/imagenet/vit/vit_base_patch16_224_b64x64_300e_jpg.py

@@ -9,6 +9,7 @@ log_config = dict(
 model = dict(
     type='Classification',
     train_preprocess=['mixUp'],
+    pretrained=False,
     mixup_cfg=dict(
         mixup_alpha=0.2,
         prob=1.0,
@@ -19,7 +20,6 @@ model = dict(
         type='PytorchImageModelWrapper',
         model_name='vit_base_patch16_224',
         num_classes=1000,
-        pretrained=False,
     ),
     head=dict(
         type='ClsHead',

configs/config_templates/classification_oss.py

@@ -29,7 +29,6 @@ model = dict(
         # model_name = 'vit_deit_small_distilled_patch16_224', # good 384,
         model_name='resnet50',
         num_classes=0,
-        pretrained=True,
     ),
     head=dict(
         type='ClsHead', with_avg_pool=True, in_channels=2048,

configs/config_templates/dino_rn50_tfrecord_oss.py

@@ -21,7 +21,7 @@ model_output_dim = 65536
 
 model = dict(
     type='DINO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=[
         'randomGrayScale', 'gaussianBlur', 'solarize'
     ],  # 2+6 view, has different augment pipeline, dino is complex
@@ -42,7 +42,6 @@ model = dict(
         # model_name = 'vit_deit_small_distilled_patch16_224', # good 384,
         model_name='resnet50',
         num_classes=0,
-        # pretrained=True,
     ),
 
     # swav need  mulit crop ,doesn't support vit based model

configs/config_templates/dino_timm.py

@@ -15,7 +15,6 @@ model_output_dim = 65536
 
 model = dict(
     type='DINO',
-    pretrained=None,
     train_preprocess=[
         'randomGrayScale', 'gaussianBlur', 'solarize'
     ],  # 2+6 view, has different augment pipeline, dino is complex
@@ -41,7 +40,6 @@ model = dict(
         # model_name = 'resnet18',
         # model_name = 'resnet34',
         # model_name = 'resnet101',
-        pretrained=True,
     ),
 
     # swav need  mulit crop ,doesn't support vit based model

configs/config_templates/dino_timm_tfrecord_oss.py

@@ -20,7 +20,7 @@ model_output_dim = 65536
 
 model = dict(
     type='DINO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=[
         'randomGrayScale', 'gaussianBlur', 'solarize'
     ],  # 2+6 view, has different augment pipeline, dino is complex
@@ -46,7 +46,6 @@ model = dict(
         # model_name = 'resnet34',
         # model_name = 'resnet101',
         # num_classes=0,
-        # pretrained=True,
     ),
 
     # swav need  mulit crop ,doesn't support vit based model

configs/config_templates/metric_learning/modelparallel_softmaxbased_tfrecord_oss.py

@@ -67,7 +67,6 @@ model = dict(
         # model_name = 'resnet34',
         # model_name = 'resnet101',
         num_classes='${backbone_channel}',
-        pretrained=True,
     ),
     neck=dict(
         type='RetrivalNeck',

configs/config_templates/metric_learning/softmaxbased_tfrecord_oss.py

@@ -72,7 +72,6 @@ model = dict(
         # model_name = 'resnet34',
         # model_name = 'resnet101',
         num_classes='${backbone_channel}',
-        pretrained=True,
     ),
     neck=dict(
         type='RetrivalNeck',

configs/config_templates/moby_timm_tfrecord_oss.py

@@ -14,7 +14,7 @@ model = dict(
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=4096,
     momentum=0.99,
-    pretrained=None,
+    pretrained=False,
     # backbone=dict(
     #     type='ResNet',
     #     depth=50,
@@ -40,7 +40,6 @@ model = dict(
         # model_name='shuffletrans_tiny_p4_w7_224', #768
         model_name='resnet50',  # 2048
         num_classes=0,
-        pretrained=False,
     ),
     neck=dict(
         type='MoBYMLP',

configs/config_templates/moco_r50_tfrecord.py

@@ -3,7 +3,7 @@ _base_ = 'configs/base.py'
 
 model = dict(
     type='MOCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=65536,
     feat_dim=128,
@@ -24,7 +24,6 @@ model = dict(
     #     #model_name = 'vit_deit_small_distilled_patch16_224', # good 384,
     #     #model_name = 'resnet50',
     #     #num_classes=1000,
-    #     pretrained=True,
     # ),
     neck=dict(
         type='NonLinearNeckV1',

configs/config_templates/moco_r50_tfrecord_oss.py

@@ -13,7 +13,7 @@ work_dir = 'oss://path/to/work_dirs/moco_r50_oss/'
 
 model = dict(
     type='MOCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=65536,
     feat_dim=128,
@@ -34,7 +34,6 @@ model = dict(
     #     #model_name = 'vit_deit_small_distilled_patch16_224', # good 384,
     #     #model_name = 'resnet50',
     #     #num_classes=1000,
-    #     pretrained=True,
     # ),
     neck=dict(
         type='NonLinearNeckV1',

configs/config_templates/moco_timm_tfrecord.py

@@ -3,7 +3,6 @@ _base_ = 'configs/base.py'
 
 model = dict(
     type='MOCO',
-    pretrained=None,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=65536,
     feat_dim=128,
@@ -21,7 +20,6 @@ model = dict(
         # model_name='xcit_medium_24_p8',  #384
         model_name='xcit_large_24_p8',  # 768
         num_classes=0,
-        pretrained=True,
     ),
     neck=dict(
         type='NonLinearNeckV1',

configs/config_templates/moco_timm_tfrecord_oss.py

@@ -13,7 +13,6 @@ work_dir = 'oss://path/to/work_dirs/moco_timm_oss/'
 
 model = dict(
     type='MOCO',
-    pretrained=None,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=65536,
     feat_dim=128,
@@ -29,7 +28,6 @@ model = dict(
         # model_name = 'resnet50',
         model_name='xcit_small_12_p8',
         num_classes=0,
-        pretrained=True,
     ),
     neck=dict(
         type='NonLinearNeckV1',

configs/config_templates/swav_r50_tfrecord.py

@@ -5,7 +5,7 @@ num_crops = [2, 6]
 # model settings
 model = dict(
     type='SWAV',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     backbone=dict(
         type='ResNet',

configs/config_templates/swav_r50_tfrecord_oss.py

@@ -15,7 +15,7 @@ num_crops = [2, 6]
 
 model = dict(
     type='SWAV',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     backbone=dict(
         type='ResNet',

configs/config_templates/topdown_litehrnet_30.py

@@ -27,7 +27,7 @@ channel_cfg = dict(
 # model settings
 model = dict(
     type='TopDown',
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='LiteHRNet',
         in_channels=3,

configs/metric_learning/imagenet_resnet50_1000kid_tfrecord.py

@@ -5,7 +5,7 @@ model = dict(
     type='Classification',
     train_preprocess=['randomErasing'],
     # train_preprocess=['mixUp'],
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='ResNet',
         depth=50,

configs/metric_learning/sop_timm_swinb_local.py

@@ -18,9 +18,7 @@ model = dict(
     'http://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/pretrained_models/timm/swin_base_patch4_window7_224_22k_statedict.pth',
     backbone=dict(
         type='PytorchImageModelWrapper',
-        model_name='swin_base_patch4_window7_224_in22k'
+        model_name='swin_base_patch4_window7_224_in22k'),
-        # pretrained=False,
-    ),
     neck=dict(
         type='RetrivalNeck',
         in_channels=1024,

configs/pose/litehrnet_30_coco_384x288.py

@@ -40,7 +40,7 @@ channel_cfg = dict(
 # model settings
 model = dict(
     type='TopDown',
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='LiteHRNet',
         in_channels=3,

configs/selfsup/byol/byol_rn50_8xb32_200e.py

@@ -3,7 +3,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='BYOL',
-    pretrained=None,
+    pretrained=False,
     base_momentum=0.996,
     backbone=dict(
         type='ResNet',

configs/selfsup/dino/dino_deit_small_p16_8xb32_100e_jpg.py

@@ -7,7 +7,7 @@ model_output_dim = 65536
 
 model = dict(
     type='DINO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=[
         'randomGrayScale', 'gaussianBlur', 'solarize'
     ],  # 2+6 view, has different augment pipeline, dino is complex
@@ -33,7 +33,6 @@ model = dict(
         # model_name = 'resnet18',
         # model_name = 'resnet34',
         # model_name = 'resnet101',
-        # pretrained=True,
     ),
 
     # swav need  mulit crop ,doesn't support vit based model

configs/selfsup/dino/dino_deit_small_p16_8xb32_100e_tfrecord.py

@@ -9,7 +9,7 @@ model_output_dim = 65536
 
 model = dict(
     type='DINO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=[
         'randomGrayScale', 'gaussianBlur', 'solarize'
     ],  # 2+6 view, has different augment pipeline, dino is complex
@@ -35,7 +35,6 @@ model = dict(
         # model_name = 'resnet34',
         # model_name = 'resnet101',
         # num_classes=0,
-        # pretrained=True,
     ),
 
     # swav need  mulit crop ,doesn't support vit based model

configs/selfsup/mixco/mixco_genet_8xb32_200e_tfrecord_oss.py

@@ -12,7 +12,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='MIXCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur', 'mixUp'],
     queue_len=65536,
     feat_dim=128,

configs/selfsup/mixco/mixco_rn50_4xb64_200e_tfrecord_oss.py

@@ -12,7 +12,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='MIXCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur', 'mixUp'],
     queue_len=65536,
     feat_dim=128,

configs/selfsup/moby/moby_deit_small_4xb32_100e_jpg.py

@@ -6,7 +6,7 @@ model = dict(
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=4096,
     momentum=0.99,
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='PytorchImageModelWrapper',
         # model_name='pit_xs_distilled_224',
@@ -16,7 +16,6 @@ model = dict(
         # model_name='vit_deit_tiny_distilled_patch16_224',   # good 192,
         model_name='vit_deit_small_distilled_patch16_224',  # good 384,
         # model_name = 'resnet50',
-        # pretrained=True,
     ),
     neck=dict(
         type='MoBYMLP',

configs/selfsup/moby/moby_deit_small_p16_4xb128_300e_tfrecord.py

@@ -15,7 +15,7 @@ model = dict(
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=4096,
     momentum=0.99,
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='PytorchImageModelWrapper',
         # model_name='pit_xs_distilled_224',
@@ -30,7 +30,6 @@ model = dict(
         # model_name='shuffletrans_tiny_p4_w7_224', #768
         # model_name = 'resnet50', # 2048
         num_classes=0,
-        pretrained=False,
     ),
     neck=dict(
         type='MoBYMLP',

configs/selfsup/moby/moby_dynamic_swin_tiny_8xb64_300e_tfrecord.py

@@ -15,12 +15,11 @@ model = dict(
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=4096,
     momentum=0.99,
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='PytorchImageModelWrapper',
         model_name='dynamic_swin_tiny_p4_w7_224',
         num_classes=0,
-        pretrained=False,
     ),
     neck=dict(
         type='MoBYMLP',

configs/selfsup/moby/moby_rn50_4xb128_100e_tfrecord.py

@@ -6,7 +6,6 @@ model = dict(
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=4096,
     momentum=0.99,
-    pretrained=None,
     # backbone=dict(
     #     type='ResNet',
     #     depth=50,
@@ -31,7 +30,6 @@ model = dict(
         # model_name='shuffletrans_tiny_p4_w7_224', #768
         model_name='resnet50',  # 2048
         num_classes=0,
-        pretrained=True,
     ),
     neck=dict(
         type='MoBYMLP',

configs/selfsup/mocov2/mocov2_rn50_8xb32_200e_jpg.py

@@ -3,7 +3,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='MOCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['gaussianBlur'],
     queue_len=65536,
     feat_dim=128,

configs/selfsup/mocov2/mocov2_rn50_8xb32_200e_tfrecord.py

@@ -3,7 +3,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='MOCO',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     queue_len=65536,
     feat_dim=128,
@@ -24,7 +24,6 @@ model = dict(
     #     #model_name = 'vit_deit_small_distilled_patch16_224', # good 384,
     #     #model_name = 'resnet50',
     #     #num_classes=1000,
-    #     pretrained=True,
     # ),
     neck=dict(
         type='NonLinearNeckV1',

configs/selfsup/simclr/simclr_rn50_8xb32_200e_jpg.py

@@ -3,7 +3,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='SimCLR',
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='ResNet',
         depth=50,

configs/selfsup/simclr/simclr_rn50_8xb32_200e_tfrecord.py

@@ -6,7 +6,7 @@ work_dir = 'work_dir/simclr/'
 # model settings
 model = dict(
     type='SimCLR',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     backbone=dict(
         type='ResNet',

configs/selfsup/simclr/simclr_rn50_mocov2_neck_8xb32_200e_jpg.py

@@ -3,7 +3,7 @@ _base_ = '../../base.py'
 # model settings
 model = dict(
     type='SimCLR',
-    pretrained=None,
+    pretrained=False,
     backbone=dict(
         type='ResNet',
         depth=50,

configs/selfsup/swav/swav_genet_8xb32_200e_tfrecord.py

@@ -6,7 +6,7 @@ num_crops = [2, 6]
 # model settings
 model = dict(
     type='SWAV',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     backbone=dict(type='PlainNet', plainnet_struct_idx='normal'),
     neck=dict(

configs/selfsup/swav/swav_rn50_8xb32_200e_tfrecord.py

@@ -5,7 +5,7 @@ num_crops = [2, 6]
 
 model = dict(
     type='SWAV',
-    pretrained=None,
+    pretrained=False,
     train_preprocess=['randomGrayScale', 'gaussianBlur'],
     backbone=dict(
         type='ResNet',

easycv/models/backbones/bninception.py

@@ -9,8 +9,6 @@ from mmcv.cnn import constant_init, kaiming_init
 from torch import nn
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import bninception as model_urls
 from ..registry import BACKBONES
 
@@ -348,6 +346,9 @@ class BNInception(nn.Module):
         if num_classes > 0:
             self.last_linear = nn.Linear(1024, num_classes)
 
+        self.default_pretrained_model_path = model_urls[
+            self.__class__.__name__]
+
     def init_weights(self):
         for m in self.modules():
             if isinstance(m, nn.Conv2d):

easycv/models/backbones/genet.py

@@ -8,8 +8,6 @@ from mmcv.cnn import constant_init, kaiming_init
 from torch import nn
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import genet as model_urls
 from ..registry import BACKBONES
 
@@ -1660,6 +1658,9 @@ class PlainNet(nn.Module):
         self.plainnet_struct = str(self) + str(self.adptive_avg_pool)
         self.zero_init_residual = False
 
+        self.default_pretrained_model_path = model_urls[self.__class__.__name__
+                                                        + plainnet_struct_idx]
+
     def init_weights(self):
         for m in self.modules():
             if isinstance(m, nn.Conv2d):

easycv/models/backbones/hrnet.py

@@ -8,8 +8,7 @@ from mmcv.cnn import (build_conv_layer, build_norm_layer, constant_init,
 from torch.nn.modules.batchnorm import _BatchNorm
 
 from easycv.models.registry import BACKBONES
-from easycv.utils.checkpoint import load_checkpoint
+from ..modelzoo import hrnet as model_urls
-from easycv.utils.logger import get_root_logger
 from .resnet import BasicBlock
 
 
@@ -592,6 +591,9 @@ class HRNet(nn.Module):
             multiscale_output=self.stage4_cfg.get('multiscale_output',
                                                   multi_scale_output))
 
+        self.default_pretrained_model_path = model_urls.get(
+            self.__class__.__name__ + arch, None)
+
     @property
     def norm1(self):
         """nn.Module: the normalization layer named "norm1" """
@@ -718,12 +720,6 @@ class HRNet(nn.Module):
         return nn.Sequential(*hr_modules), in_channels
 
     def init_weights(self):
-        """Initialize the weights in backbone.
-
-        Args:
-            pretrained (str, optional): Path to pre-trained weights.
-                Defaults to None.
-        """
         for m in self.modules():
             if isinstance(m, nn.Conv2d):
                 normal_init(m, std=0.001)

easycv/models/backbones/inceptionv3.py

@@ -11,8 +11,6 @@ import torch.nn.functional as F
 from mmcv.cnn import constant_init, kaiming_init
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import inceptionv3 as model_urls
 from ..registry import BACKBONES
 
@@ -60,6 +58,9 @@ class Inception3(nn.Module):
         if num_classes > 0:
             self.fc = nn.Linear(2048, num_classes)
 
+        self.default_pretrained_model_path = model_urls[
+            self.__class__.__name__]
+
     def init_weights(self):
         for m in self.modules():
             if isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear):

easycv/models/backbones/lighthrnet.py

@@ -12,8 +12,6 @@ from mmcv.cnn import (ConvModule, DepthwiseSeparableConvModule,
 from torch.nn.modules.batchnorm import _BatchNorm
 
 from easycv.models.registry import BACKBONES
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 
 
 def channel_shuffle(x, groups):

easycv/models/backbones/mnasnet.py

@@ -6,8 +6,6 @@ r""" This model is taken from the official PyTorch model zoo.
 import torch
 from torch import nn
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import mnasnet as model_urls
 from ..registry import BACKBONES
 
@@ -144,6 +142,9 @@ class MNASNet(torch.nn.Module):
                 nn.Dropout(p=dropout, inplace=True),
                 nn.Linear(1280, num_classes))
 
+        self.default_pretrained_model_path = model_urls[self.__class__.__name__
+                                                        + str(alpha)]
+
     def forward(self, x):
         x = self.layers(x)
         # Equivalent to global avgpool and removing H and W dimensions.

easycv/models/backbones/mobilenetv2.py

@@ -5,8 +5,6 @@ r""" This model is taken from the official PyTorch model zoo.
 
 from torch import nn
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import mobilenetv2 as model_urls
 from ..registry import BACKBONES
 
@@ -153,7 +151,8 @@ class MobileNetV2(nn.Module):
                 nn.Dropout(0.2),
                 nn.Linear(self.last_channel, num_classes),
             )
-        self.pretrained = model_urls[self.__class__.__name__ + '_' +
+        self.default_pretrained_model_path = model_urls[self.__class__.__name__
+                                                        + '_' +
                                                         str(width_multi)]
 
     def init_weights(self):

easycv/models/backbones/pytorch_image_models_wrapper.py

@@ -8,7 +8,6 @@ import torch.nn as nn
 from timm.models.helpers import load_pretrained
 from timm.models.hub import download_cached_file
 
-from easycv.utils.checkpoint import load_checkpoint
 from easycv.utils.logger import get_root_logger, print_log
 from ..modelzoo import timm_models as model_urls
 from ..registry import BACKBONES
@@ -113,35 +112,37 @@ class PytorchImageModelWrapper(nn.Module):
         logger = get_root_logger()
         if pretrained:
             if self.model_name in self.timm_model_names:
-                pretrained_path = model_urls[self.model_name]
+                default_pretrained_model_path = model_urls[self.model_name]
                 print_log(
-                    'load model from default path: {}'.format(pretrained_path),
+                    'load model from default path: {}'.format(
-                    logger)
+                        default_pretrained_model_path), logger)
-                if pretrained_path.endswith('.npz'):
+                if default_pretrained_model_path.endswith('.npz'):
                     pretrained_loc = download_cached_file(
-                        pretrained_path, check_hash=False, progress=False)
+                        default_pretrained_model_path,
+                        check_hash=False,
+                        progress=False)
                     return self.model.load_pretrained(pretrained_loc)
                 else:
                     backbone_module = importlib.import_module(
                         self.model.__module__)
                     return load_pretrained(
                         self.model,
-                        default_cfg={'url': pretrained_path},
+                        default_cfg={'url': default_pretrained_model_path},
                         filter_fn=backbone_module.checkpoint_filter_fn
                         if hasattr(backbone_module, 'checkpoint_filter_fn')
                         else None)
             elif self.model_name in _MODEL_MAP:
                 if self.model_name in model_urls.keys():
-                    pretrained_path = model_urls[self.model_name]
+                    default_pretrained_model_path = model_urls[self.model_name]
                     print_log(
                         'load model from default path: {}'.format(
-                            pretrained_path), logger)
+                            default_pretrained_model_path), logger)
                     try_max = 3
                     try_idx = 0
                     while try_idx < try_max:
                         try:
                             state_dict = torch.hub.load_state_dict_from_url(
-                                url=pretrained_path,
+                                url=default_pretrained_model_path,
                                 map_location='cpu',
                             )
                             try_idx += try_max
@@ -164,6 +165,7 @@ class PytorchImageModelWrapper(nn.Module):
                     'Error: Fail to create {} with (pretrained={}...)'.format(
                         self.model_name, pretrained))
         else:
+            print_log('load model from init weights')
             self.model.init_weights()
 
     def forward(self, x):

easycv/models/backbones/resnest.py

@@ -14,8 +14,6 @@ import torch.nn.functional as F
 from torch.nn import Conv2d, Module, ReLU
 from torch.nn.modules.utils import _pair
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..registry import BACKBONES
 
 

easycv/models/backbones/resnet.py

@@ -4,8 +4,6 @@ import torch.utils.checkpoint as cp
 from mmcv.cnn import constant_init, kaiming_init
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..modelzoo import resnet as model_urls
 from ..registry import BACKBONES
 from ..utils import FReLU, build_conv_layer, build_norm_layer
@@ -443,6 +441,9 @@ class ResNet(nn.Module):
         if num_classes > 0:
             self.fc = nn.Linear(self.feat_dim, num_classes)
 
+        self.default_pretrained_model_path = model_urls.get(
+            self.__class__.__name__ + str(depth), None)
+
     @property
     def norm1(self):
         return getattr(self, self.norm1_name)

easycv/models/backbones/resnet_jit.py

@@ -6,8 +6,6 @@ import torch.nn as nn
 from mmcv.cnn import constant_init, kaiming_init
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..registry import BACKBONES
 from ..utils import build_conv_layer, build_norm_layer
 

easycv/models/backbones/resnext.py

@@ -3,6 +3,7 @@ import math
 
 import torch.nn as nn
 
+from ..modelzoo import resnext as model_urls
 from ..registry import BACKBONES
 from ..utils import build_conv_layer, build_norm_layer
 from .resnet import Bottleneck as _Bottleneck
@@ -172,6 +173,7 @@ class ResNeXt(ResNet):
 
         self.inplanes = 64
         self.res_layers = []
+
         for i, num_blocks in enumerate(self.stage_blocks):
             stride = self.strides[i]
             dilation = self.dilations[i]
@@ -196,3 +198,7 @@ class ResNeXt(ResNet):
             self.res_layers.append(layer_name)
 
         self._freeze_stages()
+
+        self.default_pretrained_model_path = model_urls.get(
+            self.__class__.__name__ + str(self.depth) + '-' +
+            str(self.groups) + 'x' + str(self.base_width) + 'd', None)

easycv/models/backbones/shuffle_transformer.py

@@ -7,8 +7,6 @@ from einops import rearrange
 from timm.models.layers import DropPath, trunc_normal_
 from torch import nn
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..registry import BACKBONES
 
 

easycv/models/backbones/swin_transformer_dynamic.py

@@ -16,8 +16,6 @@ import torch.nn as nn
 import torch.nn.functional as F
 from timm.models.layers import DropPath, to_2tuple, trunc_normal_
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..registry import BACKBONES
 
 

easycv/models/backbones/vit_transfomer_dynamic.py

@@ -15,9 +15,6 @@ import torch.nn as nn
 # from utils import trunc_normal_
 from timm.models.layers import trunc_normal_
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
-
 
 def drop_path(x, drop_prob: float = 0., training: bool = False):
     if drop_prob == 0. or not training:

easycv/models/backbones/xcit_transformer.py

@@ -19,8 +19,6 @@ import torch.nn as nn
 from timm.models.layers import DropPath, to_2tuple, trunc_normal_
 from timm.models.vision_transformer import Mlp, _cfg
 
-from easycv.utils.checkpoint import load_checkpoint
-from easycv.utils.logger import get_root_logger
 from ..registry import BACKBONES
 
 

easycv/models/classification/classification.py

@@ -33,7 +33,7 @@ class Classification(BaseModel):
                  with_sobel=False,
                  head=None,
                  neck=None,
-                 pretrained=None,
+                 pretrained=True,
                  mixup_cfg=None):
         super(Classification, self).__init__()
         self.with_sobel = with_sobel
@@ -110,15 +110,28 @@ class Classification(BaseModel):
         self.init_weights()
 
     def init_weights(self):
-        if isinstance(self.pretrained, str):
         logger = get_root_logger()
+        if isinstance(self.pretrained, str):
             load_checkpoint(
                 self.backbone, self.pretrained, strict=False, logger=logger)
-        else:
+        elif self.pretrained:
-            print_log('load model from init weights')
             if self.backbone.__class__.__name__ == 'PytorchImageModelWrapper':
                 self.backbone.init_weights(pretrained=self.pretrained)
+            elif hasattr(self.backbone, 'default_pretrained_model_path'
+                         ) and self.backbone.default_pretrained_model_path:
+                print_log(
+                    'load model from default path: {}'.format(
+                        self.backbone.default_pretrained_model_path), logger)
+                load_checkpoint(
+                    self.backbone,
+                    self.backbone.default_pretrained_model_path,
+                    strict=False,
+                    logger=logger)
             else:
+                print_log('load model from init weights')
+                self.backbone.init_weights()
+        else:
+            print_log('load model from init weights')
             self.backbone.init_weights()
 
         for idx in range(self.head_num):

easycv/models/modelzoo.py

@@ -12,6 +12,34 @@ resnet = {
     'http://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/pretrained_models/easycv/resnet/torchvision/resnet152.pth',
 }
 
+resnext = {
+    'ResNeXt50-32x4d':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/resnext/resnext50-32x4d/epoch_100.pth',
+    'ResNeXt101-32x4d':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/resnext/resnext101-32x4d/epoch_100.pth',
+    'ResNeXt101-32x8d':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/resnext/resnext101-32x8d/epoch_100.pth',
+    'ResNext152-32x4d':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/resnext/resnext152-32x4d/epoch_100.pth',
+}
+
+hrnet = {
+    'HRNetw18':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw18/epoch_100.pth',
+    'HRNetw30':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw30/epoch_100.pth',
+    'HRNetw32':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw32/epoch_100.pth',
+    'HRNetw40':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw40/epoch_100.pth',
+    'HRNetw44':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw44/epoch_100.pth',
+    'HRNetw48':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw48/epoch_100.pth',
+    'HRNetw64':
+    'https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/EasyCV/modelzoo/classification/hrnet/hrnetw64/epoch_100.pth',
+}
+
 mobilenetv2 = {
     'MobileNetV2_1.0':
     'http://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/pretrained_models/easycv/mobilenetv2/mobilenet_v2.pth',