diff --git a/configs/_base_/datasets/ade20k.py b/configs/_base_/datasets/ade20k.py index 8f4ac8ec2..896c0196a 100644 --- a/configs/_base_/datasets/ade20k.py +++ b/configs/_base_/datasets/ade20k.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/ade20k_640x640.py b/configs/_base_/datasets/ade20k_640x640.py index a12189bf5..2e7340b38 100644 --- a/configs/_base_/datasets/ade20k_640x640.py +++ b/configs/_base_/datasets/ade20k_640x640.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2560, 640), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/chase_db1.py b/configs/_base_/datasets/chase_db1.py index ae5509022..d6bf638a5 100644 --- a/configs/_base_/datasets/chase_db1.py +++ b/configs/_base_/datasets/chase_db1.py @@ -15,6 +15,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] diff --git a/configs/_base_/datasets/cityscapes.py b/configs/_base_/datasets/cityscapes.py index 8553127a7..55d06eeda 100644 --- a/configs/_base_/datasets/cityscapes.py +++ b/configs/_base_/datasets/cityscapes.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 1024), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/cityscapes_1024x1024.py b/configs/_base_/datasets/cityscapes_1024x1024.py index 08a352723..f73af55d0 100644 --- a/configs/_base_/datasets/cityscapes_1024x1024.py +++ b/configs/_base_/datasets/cityscapes_1024x1024.py @@ -12,6 +12,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 1024), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/configs/_base_/datasets/cityscapes_768x768.py b/configs/_base_/datasets/cityscapes_768x768.py index 818644f18..b5fb57899 100644 --- a/configs/_base_/datasets/cityscapes_768x768.py +++ b/configs/_base_/datasets/cityscapes_768x768.py @@ -12,6 +12,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2049, 1025), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/configs/_base_/datasets/cityscapes_769x769.py b/configs/_base_/datasets/cityscapes_769x769.py index ed3cfef71..9a16821f5 100644 --- a/configs/_base_/datasets/cityscapes_769x769.py +++ b/configs/_base_/datasets/cityscapes_769x769.py @@ -12,6 +12,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2049, 1025), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/configs/_base_/datasets/cityscapes_832x832.py b/configs/_base_/datasets/cityscapes_832x832.py index 0f9b5815a..bbcf00c69 100644 --- a/configs/_base_/datasets/cityscapes_832x832.py +++ b/configs/_base_/datasets/cityscapes_832x832.py @@ -12,6 +12,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 1024), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/configs/_base_/datasets/coco-stuff10k.py b/configs/_base_/datasets/coco-stuff10k.py index 9cd062428..30200add3 100644 --- a/configs/_base_/datasets/coco-stuff10k.py +++ b/configs/_base_/datasets/coco-stuff10k.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/coco-stuff164k.py b/configs/_base_/datasets/coco-stuff164k.py index 099d090a4..698260b01 100644 --- a/configs/_base_/datasets/coco-stuff164k.py +++ b/configs/_base_/datasets/coco-stuff164k.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/drive.py b/configs/_base_/datasets/drive.py index 81bf858cf..87c91d5eb 100644 --- a/configs/_base_/datasets/drive.py +++ b/configs/_base_/datasets/drive.py @@ -15,6 +15,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/hrf.py b/configs/_base_/datasets/hrf.py index 6e59598c6..198bc1fcb 100644 --- a/configs/_base_/datasets/hrf.py +++ b/configs/_base_/datasets/hrf.py @@ -15,6 +15,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/isaid.py b/configs/_base_/datasets/isaid.py index e10e1b303..b57a979c2 100644 --- a/configs/_base_/datasets/isaid.py +++ b/configs/_base_/datasets/isaid.py @@ -21,6 +21,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(896, 896), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/loveda.py b/configs/_base_/datasets/loveda.py index 3bcf8bddb..167fa2d5e 100644 --- a/configs/_base_/datasets/loveda.py +++ b/configs/_base_/datasets/loveda.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(1024, 1024), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/pascal_context.py b/configs/_base_/datasets/pascal_context.py index 4b8ddc885..589006b15 100644 --- a/configs/_base_/datasets/pascal_context.py +++ b/configs/_base_/datasets/pascal_context.py @@ -17,6 +17,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/pascal_context_59.py b/configs/_base_/datasets/pascal_context_59.py index a488b83f4..85faeef47 100644 --- a/configs/_base_/datasets/pascal_context_59.py +++ b/configs/_base_/datasets/pascal_context_59.py @@ -17,6 +17,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/pascal_voc12.py b/configs/_base_/datasets/pascal_voc12.py index b423fd18f..b1aa7da7a 100644 --- a/configs/_base_/datasets/pascal_voc12.py +++ b/configs/_base_/datasets/pascal_voc12.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/potsdam.py b/configs/_base_/datasets/potsdam.py index 448e03309..4f9369849 100644 --- a/configs/_base_/datasets/potsdam.py +++ b/configs/_base_/datasets/potsdam.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(512, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/stare.py b/configs/_base_/datasets/stare.py index 09ab2e87c..adcc966c2 100644 --- a/configs/_base_/datasets/stare.py +++ b/configs/_base_/datasets/stare.py @@ -15,6 +15,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/_base_/datasets/vaihingen.py b/configs/_base_/datasets/vaihingen.py index a1eb9d984..bb7908730 100644 --- a/configs/_base_/datasets/vaihingen.py +++ b/configs/_base_/datasets/vaihingen.py @@ -14,6 +14,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(512, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/beit/upernet_beit-base_640x640_160k_ade20k_ms.py b/configs/beit/upernet_beit-base_640x640_160k_ade20k_ms.py index 81db6ac2d..323cdb13b 100644 --- a/configs/beit/upernet_beit-base_640x640_160k_ade20k_ms.py +++ b/configs/beit/upernet_beit-base_640x640_160k_ade20k_ms.py @@ -7,6 +7,9 @@ test_pipeline = [ # img_ratios is [0.5, 0.75, 1.0, 1.25, 1.5, 1.75] # original image scale is (2560, 640) dict(type='Resize', scale=(2560, 640), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs'), ] val_dataloader = dict(batch_size=1, dataset=dict(pipeline=test_pipeline)) diff --git a/configs/beit/upernet_beit-large_fp16_640x640_160k_ade20k_ms.py b/configs/beit/upernet_beit-large_fp16_640x640_160k_ade20k_ms.py index f33d534ce..279e7ace2 100644 --- a/configs/beit/upernet_beit-large_fp16_640x640_160k_ade20k_ms.py +++ b/configs/beit/upernet_beit-large_fp16_640x640_160k_ade20k_ms.py @@ -7,6 +7,9 @@ test_pipeline = [ # img_ratios is [0.5, 0.75, 1.0, 1.25, 1.5, 1.75] # original image scale is (2560, 640) dict(type='Resize', scale=(2560, 640), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs'), ] val_dataloader = dict(dataset=dict(pipeline=test_pipeline)) diff --git a/configs/cgnet/cgnet_680x680_60k_cityscapes.py b/configs/cgnet/cgnet_680x680_60k_cityscapes.py index 2995c5b09..533918829 100644 --- a/configs/cgnet/cgnet_680x680_60k_cityscapes.py +++ b/configs/cgnet/cgnet_680x680_60k_cityscapes.py @@ -38,6 +38,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 1024), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] train_dataloader = dict( diff --git a/configs/knet/knet_s3_upernet_swin-l_8x2_640x640_adamw_80k_ade20k.py b/configs/knet/knet_s3_upernet_swin-l_8x2_640x640_adamw_80k_ade20k.py index 2a52fa296..f47e33308 100644 --- a/configs/knet/knet_s3_upernet_swin-l_8x2_640x640_adamw_80k_ade20k.py +++ b/configs/knet/knet_s3_upernet_swin-l_8x2_640x640_adamw_80k_ade20k.py @@ -32,6 +32,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 640), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/configs/mae/upernet_mae-base_fp16_512x512_160k_ade20k_ms.py b/configs/mae/upernet_mae-base_fp16_512x512_160k_ade20k_ms.py index b8a9dedaf..81b913f6f 100644 --- a/configs/mae/upernet_mae-base_fp16_512x512_160k_ade20k_ms.py +++ b/configs/mae/upernet_mae-base_fp16_512x512_160k_ade20k_ms.py @@ -7,6 +7,9 @@ test_pipeline = [ # img_ratios is [0.5, 0.75, 1.0, 1.25, 1.5, 1.75] # original image scale is (2048, 512) dict(type='Resize', scale=(2048, 512), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] val_dataloader = dict(batch_size=1, dataset=dict(pipeline=test_pipeline)) diff --git a/configs/pspnet/pspnet_r50-d8_512x1024_40k_dark.py b/configs/pspnet/pspnet_r50-d8_512x1024_40k_dark.py index 93a497cff..7d82949ee 100644 --- a/configs/pspnet/pspnet_r50-d8_512x1024_40k_dark.py +++ b/configs/pspnet/pspnet_r50-d8_512x1024_40k_dark.py @@ -9,6 +9,9 @@ model = dict(preprocess_cfg=preprocess_cfg) test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(1920, 1080), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] test_dataloader = dict( diff --git a/configs/pspnet/pspnet_r50-d8_512x1024_40k_night_driving.py b/configs/pspnet/pspnet_r50-d8_512x1024_40k_night_driving.py index a26ed992e..6893efb71 100644 --- a/configs/pspnet/pspnet_r50-d8_512x1024_40k_night_driving.py +++ b/configs/pspnet/pspnet_r50-d8_512x1024_40k_night_driving.py @@ -10,6 +10,9 @@ model = dict(preprocess_cfg=preprocess_cfg) test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(1920, 1080), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] test_dataloader = dict( diff --git a/configs/pspnet/pspnet_r50-d8_512x1024_80k_dark.py b/configs/pspnet/pspnet_r50-d8_512x1024_80k_dark.py index ed6a03871..d0a3bf997 100644 --- a/configs/pspnet/pspnet_r50-d8_512x1024_80k_dark.py +++ b/configs/pspnet/pspnet_r50-d8_512x1024_80k_dark.py @@ -10,6 +10,9 @@ model = dict(preprocess_cfg=preprocess_cfg) test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(1920, 1080), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] test_dataloader = dict( diff --git a/configs/pspnet/pspnet_r50-d8_512x1024_80k_night_driving.py b/configs/pspnet/pspnet_r50-d8_512x1024_80k_night_driving.py index cec9cb6cc..c4bf6d882 100644 --- a/configs/pspnet/pspnet_r50-d8_512x1024_80k_night_driving.py +++ b/configs/pspnet/pspnet_r50-d8_512x1024_80k_night_driving.py @@ -10,6 +10,9 @@ model = dict(preprocess_cfg=preprocess_cfg) test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(1920, 1080), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations'), dict(type='PackSegInputs') ] test_dataloader = dict( diff --git a/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py b/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py index d63b7cb0d..9230ed7ad 100644 --- a/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py +++ b/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py @@ -15,6 +15,9 @@ train_pipeline = [ test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=(2048, 640), keep_ratio=True), + # add loading annotation after ``Resize`` because ground truth + # does not need to do resize data transform + dict(type='LoadAnnotations', reduce_zero_label=True), dict(type='PackSegInputs') ] train_dataloader = dict(dataset=dict(pipeline=train_pipeline)) diff --git a/tests/test_config.py b/tests/test_config.py index 41d1f9ca6..2b2bdf791 100644 --- a/tests/test_config.py +++ b/tests/test_config.py @@ -92,6 +92,8 @@ def test_config_data_pipeline(): to_float32 = load_img_pipeline.get('to_float32', False) config_mod.train_pipeline.pop(0) config_mod.test_pipeline.pop(0) + # remove loading annotation in test pipeline + config_mod.test_pipeline.pop(1) train_pipeline = Compose(config_mod.train_pipeline) test_pipeline = Compose(config_mod.test_pipeline)