From f39f02dd880bd9b5d1eeb1f15665c744c44f6945 Mon Sep 17 00:00:00 2001 From: fangyixiao18 Date: Mon, 18 Apr 2022 15:36:57 +0800 Subject: [PATCH 1/2] [Benchmark] rename linear probing config file names --- ..._in1k.py => resnet50-sobel_linear-8xb32-steplr-100e_in1k.py} | 2 +- ....py => resnet50-sobel_mhead_linear-8xb32-steplr-90e_in1k.py} | 2 +- ...b32-steplr-90e_in1k.py => resnet50_8xb32-steplr-90e_in1k.py} | 0 ...lr-100e_in1k.py => resnet50_linear-8xb32-coslr-100e_in1k.py} | 0 ...r-100e_in1k.py => resnet50_linear-8xb32-steplr-100e_in1k.py} | 0 ...slr-90e_in1k.py => resnet50_linear-8xb512-coslr-90e_in1k.py} | 0 ...e_in1k.py => resnet50_mhead_linear-8xb32-steplr-90e_in1k.py} | 0 ...r-100e_in1k.py => vit-base-p16_ft-8xb128-coslr-100e_in1k.py} | 0 ...0e_in1k.py => vit-small-p16_linear-8xb128-coslr-90e_in1k.py} | 0 9 files changed, 2 insertions(+), 2 deletions(-) rename configs/benchmarks/classification/imagenet/{resnet50-sobel_8xb32-steplr-100e_in1k.py => resnet50-sobel_linear-8xb32-steplr-100e_in1k.py} (64%) rename configs/benchmarks/classification/imagenet/{resnet50-sobel_mhead_8xb32-steplr-90e_in1k.py => resnet50-sobel_mhead_linear-8xb32-steplr-90e_in1k.py} (62%) rename configs/benchmarks/classification/imagenet/{resnet50-nofrz_8xb32-steplr-90e_in1k.py => resnet50_8xb32-steplr-90e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{resnet50_8xb32-coslr-100e_in1k.py => resnet50_linear-8xb32-coslr-100e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{resnet50_8xb32-steplr-100e_in1k.py => resnet50_linear-8xb32-steplr-100e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{resnet50_8xb512-coslr-90e_in1k.py => resnet50_linear-8xb512-coslr-90e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{resnet50_mhead_8xb32-steplr-90e_in1k.py => resnet50_mhead_linear-8xb32-steplr-90e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{vit-b-p16_ft-8xb128-coslr-100e_in1k.py => vit-base-p16_ft-8xb128-coslr-100e_in1k.py} (100%) rename configs/benchmarks/classification/imagenet/{vit-small-p16_8xb128-coslr-90e_in1k.py => vit-small-p16_linear-8xb128-coslr-90e_in1k.py} (100%) diff --git a/configs/benchmarks/classification/imagenet/resnet50-sobel_8xb32-steplr-100e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50-sobel_linear-8xb32-steplr-100e_in1k.py similarity index 64% rename from configs/benchmarks/classification/imagenet/resnet50-sobel_8xb32-steplr-100e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50-sobel_linear-8xb32-steplr-100e_in1k.py index d0f759291..f8fb3b2df 100644 --- a/configs/benchmarks/classification/imagenet/resnet50-sobel_8xb32-steplr-100e_in1k.py +++ b/configs/benchmarks/classification/imagenet/resnet50-sobel_linear-8xb32-steplr-100e_in1k.py @@ -1,4 +1,4 @@ -_base_ = 'resnet50_8xb32-steplr-100e_in1k.py' +_base_ = 'resnet50_linear-8xb32-steplr-100e_in1k.py' # model settings model = dict(with_sobel=True, backbone=dict(in_channels=2, frozen_stages=4)) diff --git a/configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_8xb32-steplr-90e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_linear-8xb32-steplr-90e_in1k.py similarity index 62% rename from configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_8xb32-steplr-90e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_linear-8xb32-steplr-90e_in1k.py index 5047ac10c..37a434185 100644 --- a/configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_8xb32-steplr-90e_in1k.py +++ b/configs/benchmarks/classification/imagenet/resnet50-sobel_mhead_linear-8xb32-steplr-90e_in1k.py @@ -1,4 +1,4 @@ -_base_ = 'resnet50_mhead_8xb32-steplr-90e_in1k.py' +_base_ = 'resnet50_mhead_linear-8xb32-steplr-90e_in1k.py' # model settings model = dict(with_sobel=True, backbone=dict(in_channels=2, frozen_stages=4)) diff --git a/configs/benchmarks/classification/imagenet/resnet50-nofrz_8xb32-steplr-90e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-90e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/resnet50-nofrz_8xb32-steplr-90e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-90e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/resnet50_8xb32-coslr-100e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-coslr-100e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/resnet50_8xb32-coslr-100e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-coslr-100e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/resnet50_8xb512-coslr-90e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50_linear-8xb512-coslr-90e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/resnet50_8xb512-coslr-90e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50_linear-8xb512-coslr-90e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py b/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py rename to configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/vit-b-p16_ft-8xb128-coslr-100e_in1k.py b/configs/benchmarks/classification/imagenet/vit-base-p16_ft-8xb128-coslr-100e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/vit-b-p16_ft-8xb128-coslr-100e_in1k.py rename to configs/benchmarks/classification/imagenet/vit-base-p16_ft-8xb128-coslr-100e_in1k.py diff --git a/configs/benchmarks/classification/imagenet/vit-small-p16_8xb128-coslr-90e_in1k.py b/configs/benchmarks/classification/imagenet/vit-small-p16_linear-8xb128-coslr-90e_in1k.py similarity index 100% rename from configs/benchmarks/classification/imagenet/vit-small-p16_8xb128-coslr-90e_in1k.py rename to configs/benchmarks/classification/imagenet/vit-small-p16_linear-8xb128-coslr-90e_in1k.py From 6483ae7a979ce7ea9b40632f65c839367b8e0056 Mon Sep 17 00:00:00 2001 From: fangyixiao18 Date: Mon, 25 Apr 2022 11:02:09 +0800 Subject: [PATCH 2/2] update config links --- configs/selfsup/byol/README.md | 4 ++-- configs/selfsup/deepcluster/README.md | 4 ++-- configs/selfsup/densecl/README.md | 4 ++-- configs/selfsup/mae/README.md | 6 +++--- configs/selfsup/mocov2/README.md | 4 ++-- configs/selfsup/mocov3/README.md | 6 +++--- configs/selfsup/npid/README.md | 4 ++-- configs/selfsup/odc/README.md | 4 ++-- configs/selfsup/relative_loc/README.md | 4 ++-- configs/selfsup/rotation_pred/README.md | 4 ++-- configs/selfsup/simclr/README.md | 4 ++-- configs/selfsup/simsiam/README.md | 4 ++-- configs/selfsup/swav/README.md | 4 ++-- demo/mmselfsup_colab_tutorial.ipynb | 2 +- 14 files changed, 29 insertions(+), 29 deletions(-) diff --git a/configs/selfsup/byol/README.md b/configs/selfsup/byol/README.md index 988972b2d..9d50bb8d9 100644 --- a/configs/selfsup/byol/README.md +++ b/configs/selfsup/byol/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/deepcluster/README.md b/configs/selfsup/deepcluster/README.md index f11f32a54..a6a6f907e 100644 --- a/configs/selfsup/deepcluster/README.md +++ b/configs/selfsup/deepcluster/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/densecl/README.md b/configs/selfsup/densecl/README.md index ed12476d5..0ff5f4558 100644 --- a/configs/selfsup/densecl/README.md +++ b/configs/selfsup/densecl/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | -------------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/mae/README.md b/configs/selfsup/mae/README.md index 3831bee9b..230ee30b1 100644 --- a/configs/selfsup/mae/README.md +++ b/configs/selfsup/mae/README.md @@ -35,9 +35,9 @@ for 400 epochs, the details are below: -| Backbone | Pre-train epoch | Fine-tuning Top-1 | Pre-train Config | Fine-tuning Config | Download | -| :------: | :-------------: | :---------------: | :-------------------------------------------------: | :---------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | -| ViT-B/16 | 400 | 83.1 | [config](https://github.com/open-mmlab/mmselfsup/blob/master/configs/selfsup/mae/mae_vit-b-p16_8xb512-coslr-400e_in1k.py) | [config](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/vit-b-p16_ft-8xb128-coslr-100e_in1k.py) | [model](https://download.openmmlab.com/mmselfsup/mae/mae_vit-base-p16_8xb512-coslr-400e_in1k-224_20220223-85be947b.pth) | [log](https://download.openmmlab.com/mmselfsup/mae/mae_vit-base-p16_8xb512-coslr-300e_in1k-224_20220210_140925.log.json) | +| Backbone | Pre-train epoch | Fine-tuning Top-1 | Pre-train Config | Fine-tuning Config | Download | +| :------: | :-------------: | :---------------: | :-----------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | +| ViT-B/16 | 400 | 83.1 | [config](https://github.com/open-mmlab/mmselfsup/blob/master/configs/selfsup/mae/mae_vit-b-p16_8xb512-coslr-400e_in1k.py) | [config](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/vit-base-p16_ft-8xb128-coslr-100e_in1k.py) | [model](https://download.openmmlab.com/mmselfsup/mae/mae_vit-base-p16_8xb512-coslr-400e_in1k-224_20220223-85be947b.pth) | [log](https://download.openmmlab.com/mmselfsup/mae/mae_vit-base-p16_8xb512-coslr-300e_in1k-224_20220210_140925.log.json) | ## Citation diff --git a/configs/selfsup/mocov2/README.md b/configs/selfsup/mocov2/README.md index 7014b5a2f..96ce39d75 100644 --- a/configs/selfsup/mocov2/README.md +++ b/configs/selfsup/mocov2/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ------------------------------------------------------------------------------------------------------------------------------------------------ | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/mocov3/README.md b/configs/selfsup/mocov3/README.md index e80646d57..ef9043dee 100644 --- a/configs/selfsup/mocov3/README.md +++ b/configs/selfsup/mocov3/README.md @@ -26,9 +26,9 @@ The classification benchmarks includes 4 downstream task datasets, **VOC**, **Im The **Linear Evaluation** result is obtained by training a linear head upon the pre-trained backbone. Please refer to [vit-small-p16_8xb128-coslr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/vit-small-p16_8xb128-coslr-90e_in1k.py) for details of config. -| Self-Supervised Config | Linear Evaluation | -| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------- | -| [vit-small-p16_32xb128-fp16-coslr-300e_in1k-224](https://github.com/open-mmlab/mmselfsup/blob/master/configs/selfsup/mocov3/mocov3_vit-small-p16_32xb128-fp16-coslr-300e_in1k-224.py) | 73.19 | +| Self-Supervised Config | Linear Evaluation | +| --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------- | +| [vit-small-p16_linear-32xb128-fp16-coslr-300e_in1k-224](https://github.com/open-mmlab/mmselfsup/blob/master/configs/selfsup/mocov3/mocov3_vit-small-p16_linear-32xb128-fp16-coslr-300e_in1k-224.py) | 73.19 | ## Citation diff --git a/configs/selfsup/npid/README.md b/configs/selfsup/npid/README.md index 3e9462b56..d7f561054 100644 --- a/configs/selfsup/npid/README.md +++ b/configs/selfsup/npid/README.md @@ -38,9 +38,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ---------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/odc/README.md b/configs/selfsup/odc/README.md index dc1d9b597..ba29f8444 100644 --- a/configs/selfsup/odc/README.md +++ b/configs/selfsup/odc/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | -------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/relative_loc/README.md b/configs/selfsup/relative_loc/README.md index 9529647ad..5f8dc1c56 100644 --- a/configs/selfsup/relative_loc/README.md +++ b/configs/selfsup/relative_loc/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/rotation_pred/README.md b/configs/selfsup/rotation_pred/README.md index 97e9d5a97..ffade3016 100644 --- a/configs/selfsup/rotation_pred/README.md +++ b/configs/selfsup/rotation_pred/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/simclr/README.md b/configs/selfsup/simclr/README.md index 93be60128..d3ed630b2 100644 --- a/configs/selfsup/simclr/README.md +++ b/configs/selfsup/simclr/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-steplr-100e_in1k.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-steplr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ------------------------------------------------------------------------------------------------------------------------------------------------ | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/simsiam/README.md b/configs/selfsup/simsiam/README.md index 2d4f9356e..a710a0721 100644 --- a/configs/selfsup/simsiam/README.md +++ b/configs/selfsup/simsiam/README.md @@ -35,9 +35,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb512-coslr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb512-coslr-90e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb512-coslr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb512-coslr-90e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | -------------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/configs/selfsup/swav/README.md b/configs/selfsup/swav/README.md index bf289d99d..51afc1074 100644 --- a/configs/selfsup/swav/README.md +++ b/configs/selfsup/swav/README.md @@ -34,9 +34,9 @@ Besides, k=1 to 96 indicates the hyper-parameter of Low-shot SVM. #### ImageNet Linear Evaluation -The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_8xb32-steplr-90e.py](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_8xb32-steplr-90e_in1k.py) for details of config. +The **Feature1 - Feature5** don't have the GlobalAveragePooling, the feature map is pooled to the specific dimensions and then follows a Linear layer to do the classification. Please refer to [resnet50_mhead_linear-8xb32-steplr-90e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_mhead_linear-8xb32-steplr-90e_in1k.py) for details of config. -The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_8xb32-coslr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_8xb32-coslr-100e_in1k.py) for details of config. +The **AvgPool** result is obtained from Linear Evaluation with GlobalAveragePooling. Please refer to [resnet50_linear-8xb32-coslr-100e_in1k](https://github.com/open-mmlab/mmselfsup/blob/master/configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-coslr-100e_in1k.py) for details of config. | Self-Supervised Config | Feature1 | Feature2 | Feature3 | Feature4 | Feature5 | AvgPool | | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------- | -------- | -------- | -------- | -------- | ------- | diff --git a/demo/mmselfsup_colab_tutorial.ipynb b/demo/mmselfsup_colab_tutorial.ipynb index 0fcb6fa2f..75af94202 100644 --- a/demo/mmselfsup_colab_tutorial.ipynb +++ b/demo/mmselfsup_colab_tutorial.ipynb @@ -1547,7 +1547,7 @@ "source": [ "# Load the basic config file\n", "from mmcv import Config\n", - "benchmark_cfg = Config.fromfile('configs/benchmarks/classification/imagenet/resnet50_8xb32-steplr-100e_in1k.py')\n", + "benchmark_cfg = Config.fromfile('configs/benchmarks/classification/imagenet/resnet50_linear-8xb32-steplr-100e_in1k.py')\n", "\n", "# Modify the model\n", "checkpoint_file = 'work_dirs/selfsup/relative-loc_resnet50_8xb64-steplr-70e_in1k_colab/relative-loc_backbone-weights.pth'\n",