Bump version to v0.2.6 #20

Westlake-AI · Sep 14, 2022 · 7785c01 · 7785c01
1 parent ab97e2a
commit 7785c01
Show file tree

Hide file tree

Showing 83 changed files with 4,584 additions and 68 deletions.
diff --git a/README.md b/README.md
@@ -41,9 +41,7 @@ The main branch works with **PyTorch 1.8** (required by some self-supervised met
 
 ## What's New
 
-[2020-08-19] Weights and logs for mixup benchmarks are released.
-
-[2020-07-30] `OpenMixup` v0.2.5 is released (issue [#10](https://github.com/Westlake-AI/openmixup/issues/10)).
+[2020-09-14] `OpenMixup` v0.2.6 is released (issue [#20](https://github.com/Westlake-AI/openmixup/issues/20)).
 
 ## Installation
 
@@ -91,6 +89,7 @@ Please refer to [Model Zoos](docs/en/model_zoos) for various backbones, mixup me
     - [x] [ResNeXt](https://arxiv.org/abs/1611.05431) (CVPR'2017) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/resnet/)]
     - [x] [SE-ResNet](https://arxiv.org/abs/1709.01507) (CVPR'2018) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/resnet/)]
     - [x] [SE-ResNeXt](https://arxiv.org/abs/1709.01507) (CVPR'2018) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/resnet/)]
+    - [x] [ShuffleNetV1](https://arxiv.org/abs/1807.11164) (CVPR'2018)
     - [x] [ShuffleNetV2](https://arxiv.org/abs/1807.11164) (ECCV'2018) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/shufflenet/)]
     - [x] [MobileNetV2](https://arxiv.org/abs/1801.04381) (CVPR'2018) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/mobilenet_v2/)]
     - [x] [MobileNetV3](https://arxiv.org/abs/1905.02244) (ICCV'2019)
@@ -104,8 +103,12 @@ Please refer to [Model Zoos](docs/en/model_zoos) for various backbones, mixup me
     - [x] [UniFormer](https://arxiv.org/abs/2201.09450) (ICLR'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/uniformer/)]
     - [x] [PoolFormer](https://arxiv.org/abs/2111.11418) (CVPR'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/poolformer/)]
     - [x] [ConvNeXt](https://arxiv.org/abs/2201.03545) (CVPR'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/convnext/)]
+    - [x] [MViTV2](https://arxiv.org/abs/2112.01526) (CVPR'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/mvit/)]
     - [x] [VAN](https://arxiv.org/abs/2202.09741) (ArXiv'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/van/)]
     - [x] [LITv2](https://arxiv.org/abs/2205.13213) (ArXiv'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/lit_v2/)]
+    - [x] [HorNet](https://arxiv.org/abs/2207.14284) (ECCV'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/hornet/)]
+    - [x] [EdgeNeXt](https://arxiv.org/abs/2206.10589) (ECCVW'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/edgenext/)]
+    - [x] [EfficientFormer](https://arxiv.org/abs/2206.01191) (ArXiv'2022) [[config](https://github.com/Westlake-AI/openmixup/tree/main/configs/classification/imagenet/efficientformer/)]
     </details>
 
 * Mixup methods for supervised image classification.
@@ -182,22 +185,32 @@ This project is released under the [Apache 2.0 license](LICENSE).
 
 ## Citation
 
-If you find this project useful in your research, please consider cite our repo:
+If you find this project useful in your research, please consider cite our repo and tech report:
 
 ```BibTeX
 @misc{2022openmixup,
     title={{OpenMixup}: Open Mixup Toolbox and Benchmark for Visual Representation Learning},
-    author={Li, Siyuan and Liu, Zicheng and Wu, Di and Stan Z. Li},
+    author={Li, Siyuan and Liu, Zicheng and Wang, Zedong and Wu, Di and Stan Z. Li},
     publisher = {GitHub},
     journal = {GitHub repository},
     howpublished = {\url{https://github.com/Westlake-AI/openmixup}},
     year={2022}
 }
 ```
+```BibTeX
+@misc{li2022openmixup,
+    title={OpenMixup: Open Mixup Toolbox and Benchmark for Visual Representation Learning}, 
+    author={Li, Siyuan and Wang, Zedong and Liu, Zicheng and Wu, Di and Stan Z. Li},
+    year={2022},
+    eprint={2209.04851},
+    archivePrefix={arXiv},
+    primaryClass={cs.CV}
+}
+```
 
 ## Contributors
 
-For now, the direct contributors include: Siyuan Li ([@Lupin1998](https://github.com/Lupin1998)), Zicheng Liu ([@pone7](https://github.com/pone7)), Zedong Wang ([@Jacky1128](https://github.com/Jacky1128)), and Di Wu ([@wudi-bu](https://github.com/wudi-bu)). We thank contributors from MMSelfSup and MMClassification.
+For now, the direct contributors include: Siyuan Li ([@Lupin1998](https://github.com/Lupin1998)), Zicheng Liu ([@pone7](https://github.com/pone7)), Zedong Wang ([@Jacky1128](https://github.com/Jacky1128)), and Di Wu ([@wudi-bu](https://github.com/wudi-bu)). We thank contributors from MMSelfSup and MMClassification and all public contributors!
 
 ## Contact
 

diff --git a/configs/classification/_base_/datasets/imagenet/edgenext_sz256_8xbs128.py b/configs/classification/_base_/datasets/imagenet/edgenext_sz256_8xbs128.py
@@ -0,0 +1,89 @@
+# Refers to `_RAND_INCREASING_TRANSFORMS` in pytorch-image-models
+rand_increasing_policies = [
+    dict(type='AutoContrast'),
+    dict(type='Equalize'),
+    dict(type='Invert'),
+    dict(type='Rotate', magnitude_key='angle', magnitude_range=(0, 30)),
+    dict(type='Posterize', magnitude_key='bits', magnitude_range=(4, 0)),
+    dict(type='Solarize', magnitude_key='thr', magnitude_range=(256, 0)),
+    dict(type='SolarizeAdd', magnitude_key='magnitude', magnitude_range=(0, 110)),
+    dict(type='ColorTransform', magnitude_key='magnitude', magnitude_range=(0, 0.9)),
+    dict(type='Contrast', magnitude_key='magnitude', magnitude_range=(0, 0.9)),
+    dict(type='Brightness', magnitude_key='magnitude', magnitude_range=(0, 0.9)),
+    dict(type='Sharpness', magnitude_key='magnitude', magnitude_range=(0, 0.9)),
+    dict(type='Shear',
+        magnitude_key='magnitude', magnitude_range=(0, 0.3), direction='horizontal'),
+    dict(type='Shear',
+        magnitude_key='magnitude', magnitude_range=(0, 0.3), direction='vertical'),
+    dict(type='Translate',
+        magnitude_key='magnitude', magnitude_range=(0, 0.45), direction='horizontal'),
+    dict(type='Translate',
+        magnitude_key='magnitude', magnitude_range=(0, 0.45), direction='vertical'),
+]
+
+# dataset settings
+data_source_cfg = dict(type='ImageNet')
+# ImageNet dataset
+data_train_list = 'data/meta/ImageNet/train_labeled_full.txt'
+data_train_root = 'data/ImageNet/train'
+data_test_list = 'data/meta/ImageNet/val_labeled.txt'
+data_test_root = 'data/ImageNet/val/'
+
+dataset_type = 'ClassificationDataset'
+img_norm_cfg = dict(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+train_pipeline = [
+    dict(type='RandomResizedCrop', size=256, interpolation=3),  # bicubic
+    dict(type='RandomHorizontalFlip'),
+    dict(type='RandAugment',
+        policies=rand_increasing_policies,
+        num_policies=2, total_level=10,
+        magnitude_level=9, magnitude_std=0.5,
+        hparams=dict(
+            pad_val=[104, 116, 124], interpolation='bicubic')),
+    dict(type='ColorJitter', brightness=0.4, contrast=0.4, saturation=0.4),
+    dict(
+        type='RandomErasing_numpy',  # before ToTensor and Normalize
+        erase_prob=0.25,
+        mode='rand', min_area_ratio=0.02, max_area_ratio=1 / 3,
+        fill_color=[104, 116, 124], fill_std=[58, 57, 57]),  # RGB
+]
+test_pipeline = [
+    dict(type='Resize', size=292, interpolation=3),  # 0.875
+    dict(type='CenterCrop', size=256),
+    dict(type='ToTensor'),
+    dict(type='Normalize', **img_norm_cfg),
+]
+# prefetch
+prefetch = True
+if not prefetch:
+    train_pipeline.extend([dict(type='ToTensor'), dict(type='Normalize', **img_norm_cfg)])
+
+data = dict(
+    imgs_per_gpu=128,
+    workers_per_gpu=8,
+    train=dict(
+        type=dataset_type,
+        data_source=dict(
+            list_file=data_train_list, root=data_train_root,
+            **data_source_cfg),
+        pipeline=train_pipeline,
+        prefetch=prefetch,
+    ),
+    val=dict(
+        type=dataset_type,
+        data_source=dict(
+            list_file=data_test_list, root=data_test_root, **data_source_cfg),
+        pipeline=test_pipeline,
+        prefetch=False,
+    ))
+
+# validation hook
+evaluation = dict(
+    initial=False,
+    interval=1,
+    imgs_per_gpu=128,
+    workers_per_gpu=4,
+    eval_param=dict(topk=(1, 5)))
+
+# checkpoint
+checkpoint_config = dict(interval=1, max_keep_ckpts=1)
diff --git a/configs/classification/_base_/models/edgenext/edgenext-base.py b/configs/classification/_base_/models/edgenext/edgenext-base.py
@@ -0,0 +1,25 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EdgeNeXt',
+        arch='base',
+        out_indices=(3, ),
+        drop_path_rate=0.1,
+        gap_before_final_norm=True,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=584, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='BatchNorm', val=1., bias=0.)
+    ],
+)
diff --git a/configs/classification/_base_/models/edgenext/edgenext-small.py b/configs/classification/_base_/models/edgenext/edgenext-small.py
@@ -0,0 +1,25 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EdgeNeXt',
+        arch='small',
+        out_indices=(3, ),
+        drop_path_rate=0.1,
+        gap_before_final_norm=True,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=304, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='BatchNorm', val=1., bias=0.)
+    ],
+)
diff --git a/configs/classification/_base_/models/edgenext/edgenext-xsmall.py b/configs/classification/_base_/models/edgenext/edgenext-xsmall.py
@@ -0,0 +1,25 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EdgeNeXt',
+        arch='xsmall',
+        out_indices=(3, ),
+        drop_path_rate=0.1,
+        gap_before_final_norm=True,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=192, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='BatchNorm', val=1., bias=0.)
+    ],
+)
diff --git a/configs/classification/_base_/models/edgenext/edgenext-xxsmall.py b/configs/classification/_base_/models/edgenext/edgenext-xxsmall.py
@@ -0,0 +1,25 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EdgeNeXt',
+        arch='xxsmall',
+        out_indices=(3, ),
+        drop_path_rate=0.1,
+        gap_before_final_norm=True,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=168, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='BatchNorm', val=1., bias=0.)
+    ],
+)
diff --git a/configs/classification/_base_/models/efficientformer/efficientformer_l1.py b/configs/classification/_base_/models/efficientformer/efficientformer_l1.py
@@ -0,0 +1,24 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EfficientFormer',
+        arch='l1',
+        drop_path_rate=0.,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=448, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer=['BatchNorm', 'LayerNorm'], val=1., bias=0.),
+        dict(type='Constant', layer=['LayerScale'], val=1e-5),
+    ],
+)
diff --git a/configs/classification/_base_/models/efficientformer/efficientformer_l3.py b/configs/classification/_base_/models/efficientformer/efficientformer_l3.py
@@ -0,0 +1,24 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EfficientFormer',
+        arch='l3',
+        drop_path_rate=0.,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=512, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer=['BatchNorm', 'LayerNorm'], val=1., bias=0.),
+        dict(type='Constant', layer=['LayerScale'], val=1e-5),
+    ],
+)
diff --git a/configs/classification/_base_/models/efficientformer/efficientformer_l7.py b/configs/classification/_base_/models/efficientformer/efficientformer_l7.py
@@ -0,0 +1,24 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='EfficientFormer',
+        arch='l7',
+        drop_path_rate=0.,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=768, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Conv2d', 'Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer=['BatchNorm', 'LayerNorm'], val=1., bias=0.),
+        dict(type='Constant', layer=['LayerScale'], val=1e-5),
+    ],
+)
diff --git a/configs/classification/_base_/models/hornet/hornet_base.py b/configs/classification/_base_/models/hornet/hornet_base.py
@@ -0,0 +1,24 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='HorNet',
+        arch='base',
+        drop_path_rate=0.5,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=1024, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='LayerNorm', val=1., bias=0.),
+        dict(type='Constant', layer=['LayerScale'], val=1e-6),
+    ],
+)
diff --git a/configs/classification/_base_/models/hornet/hornet_large.py b/configs/classification/_base_/models/hornet/hornet_large.py
@@ -0,0 +1,24 @@
+# model settings
+model = dict(
+    type='MixUpClassification',
+    pretrained=None,
+    alpha=[0.8, 1.0,],
+    mix_mode=["mixup", "cutmix",],
+    mix_args=dict(),
+    backbone=dict(
+        type='HorNet',
+        arch='large',
+        drop_path_rate=0.2,
+    ),
+    head=dict(
+        type='ClsMixupHead',  # mixup CE + label smooth
+        loss=dict(type='LabelSmoothLoss',
+            label_smooth_val=0.1, num_classes=1000, mode='original', loss_weight=1.0),
+        with_avg_pool=True,
+        in_channels=1536, num_classes=1000),
+    init_cfg=[
+        dict(type='TruncNormal', layer=['Linear'], std=0.02, bias=0.),
+        dict(type='Constant', layer='LayerNorm', val=1., bias=0.),
+        dict(type='Constant', layer=['LayerScale'], val=1e-6),
+    ],
+)