diff --git a/configs/mobileone/mobileone-s0_8xb128_in1k.py b/configs/_base_/schedules/imagenet_bs256_coslr_coswd_300e.py similarity index 63% rename from configs/mobileone/mobileone-s0_8xb128_in1k.py rename to configs/_base_/schedules/imagenet_bs256_coslr_coswd_300e.py index ceeb21b7fe2..318e0315743 100644 --- a/configs/mobileone/mobileone-s0_8xb128_in1k.py +++ b/configs/_base_/schedules/imagenet_bs256_coslr_coswd_300e.py @@ -1,19 +1,6 @@ -_base_ = [ - '../_base_/models/mobileone/mobileone_s0.py', - '../_base_/datasets/imagenet_bs32_pil_resize.py', - '../_base_/default_runtime.py' -] - -# dataset settings -train_dataloader = dict(batch_size=128) -val_dataloader = dict(batch_size=128) -test_dataloader = dict(batch_size=128) - -# schedule settings +# optimizer optim_wrapper = dict( - optimizer=dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001), - paramwise_cfg=dict(bias_decay_mult=0., norm_decay_mult=0.), -) + optimizer=dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)) # learning policy param_scheduler = [ @@ -50,7 +37,4 @@ # NOTE: `auto_scale_lr` is for automatically scaling LR, # based on the actual training batch size. -auto_scale_lr = dict(base_batch_size=1024) - -# runtime setting -custom_hooks = [dict(type='EMAHook', momentum=5e-4, priority='ABOVE_NORMAL')] +auto_scale_lr = dict(base_batch_size=256) diff --git a/configs/mobileone/README.md b/configs/mobileone/README.md index 80fbb148359..ce187e262d5 100644 --- a/configs/mobileone/README.md +++ b/configs/mobileone/README.md @@ -4,122 +4,163 @@ -## Abstract +## Introduction -Efficient neural network backbones for mobile devices are often optimized for metrics such as FLOPs or parameter count. However, these metrics may not correlate well with latency of the network when deployed on a mobile device. Therefore, we perform extensive analysis of different metrics by deploying several mobile-friendly networks on a mobile device. We identify and analyze architectural and optimization bottlenecks in recent efficient neural networks and provide ways to mitigate these bottlenecks. To this end, we design an efficient backbone MobileOne, with variants achieving an inference time under 1 ms on an iPhone12 with 75.9% top-1 accuracy on ImageNet. We show that MobileOne achieves state-of-the-art performance within the efficient architectures while being many times faster on mobile. Our best model obtains similar performance on ImageNet as MobileFormer while being 38x faster. Our model obtains 2.3% better top-1 accuracy on ImageNet than EfficientNet at similar latency. Furthermore, we show that our model generalizes to multiple tasks - image classification, object detection, and semantic segmentation with significant improvements in latency and accuracy as compared to existing efficient architectures when deployed on a mobile device. +Mobileone is proposed by apple and based on reparameterization. On the apple chips, the accuracy of the model is close to 0.76 on the ImageNet dataset when the latency is less than 1ms. Its main improvements based on [RepVGG](../repvgg) are fllowing: + +- Reparameterization using Depthwise convolution and Pointwise convolution instead of normal convolution. +- Removal of the residual structure which is not friendly to access memory.
-## Results and models +## Abstract -### ImageNet-1k +
-| Model | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download | -| :------------: | :-----------------------------: | :----------------------------: | :-------: | :-------: | :--------------------------------------------------: | :-----------------------------------------------------: | -| MobileOne-s0\* | 5.29（train) \| 2.08 (deploy) | 1.09 (train) \| 0.28 (deploy) | 71.36 | 89.87 | [config (train)](./mobileone-s0_8xb128_in1k.py) \| [config (deploy)](./deploy/mobileone-s0_deploy_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_3rdparty_in1k_20220915-007ae971.pth) | -| MobileOne-s1\* | 4.83 (train) \| 4.76 (deploy) | 0.86 (train) \| 0.84 (deploy) | 75.76 | 92.77 | [config (train)](./mobileone-s1_8xb128_in1k.py) \| [config (deploy)](./deploy/mobileone-s1_deploy_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s1_3rdparty_in1k_20220915-473c8469.pth) | -| MobileOne-s2\* | 7.88 (train) \| 7.88 (deploy) | 1.34 (train) \| 1.31 (deploy) | 77.39 | 93.63 | [config (train)](./mobileone-s2_8xb128_in1k.py) \|[config (deploy)](./deploy/mobileone-s2_deploy_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s2_3rdparty_in1k_20220915-ed2e4c30.pth) | -| MobileOne-s3\* | 10.17 (train) \| 10.08 (deploy) | 1.95 (train) \| 1.91 (deploy) | 77.93 | 93.89 | [config (train)](./mobileone-s3_8xb128_in1k.py) \|[config (deploy)](./deploy/mobileone-s3_deploy_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s3_3rdparty_in1k_20220915-84d6a02c.pth) | -| MobileOne-s4\* | 14.95 (train) \| 14.84 (deploy) | 3.05 (train) \| 3.00 (deploy) | 79.30 | 94.37 | [config (train)](./mobileone-s4_8xb128_in1k.py) \|[config (deploy)](./deploy/mobileone-s4_deploy_8xb128_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s4_3rdparty_in1k_20220915-ce9509ee.pth) | +Show the paper's abstract -*Models with * are converted from the [official repo](https://github.com/apple/ml-mobileone). The config files of these models are only for validation. We don't ensure these config files' training accuracy and welcome you to contribute your reproduction results.* +
+Efficient neural network backbones for mobile devices are often optimized for metrics such as FLOPs or parameter count. However, these metrics may not correlate well with latency of the network when deployed on a mobile device. Therefore, we perform extensive analysis of different metrics by deploying several mobile-friendly networks on a mobile device. We identify and analyze architectural and optimization bottlenecks in recent efficient neural networks and provide ways to mitigate these bottlenecks. To this end, we design an efficient backbone MobileOne, with variants achieving an inference time under 1 ms on an iPhone12 with 75.9% top-1 accuracy on ImageNet. We show that MobileOne achieves state-of-the-art performance within the efficient architectures while being many times faster on mobile. Our best model obtains similar performance on ImageNet as MobileFormer while being 38x faster. Our model obtains 2.3% better top-1 accuracy on ImageNet than EfficientNet at similar latency. Furthermore, we show that our model generalizes to multiple tasks - image classification, object detection, and semantic segmentation with significant improvements in latency and accuracy as compared to existing efficient architectures when deployed on a mobile device. +
-*Because the [official repo.](https://github.com/apple/ml-mobileone) does not give a strategy for training and testing, the test data pipline of [RepVGG](https://github.com/open-mmlab/mmclassification/tree/master/configs/repvgg) is used here, and the result is about 0.1 lower than that in the paper. Refer to [this issue](https://github.com/apple/ml-mobileone/issues/2).* +
## How to use -The checkpoints provided are all `training-time` models. Use the reparameterize tool to switch them to more efficient `inference-time` architecture, which not only has fewer parameters but also less calculations. +The checkpoints provided are all `training-time` models. Use the reparameterize tool or `switch_to_deploy` interface to switch them to more efficient `inference-time` architecture, which not only has fewer parameters but also less calculations. -### Use tool + -Use provided tool to reparameterize the given model and save the checkpoint: +**Predict image** -```bash -python tools/convert_models/reparameterize_model.py ${CFG_PATH} ${SRC_CKPT_PATH} ${TARGET_CKPT_PATH} +Use `classifier.backbone.switch_to_deploy()` interface to switch the MobileOne to a inference mode. + +```python +>>> import torch +>>> from mmcls.apis import init_model, inference_model +>>> +>>> model = init_model('configs/mobileone/mobileone-s0_8xb32_in1k.py', 'https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth') +>>> predict = inference_model(model, 'demo/demo.JPEG') +>>> print(predict['pred_class']) +sea snake +>>> print(predict['pred_score']) +0.4539405107498169 +>>> +>>> # switch to deploy mode +>>> model.backbone.switch_to_deploy() +>>> predict_deploy = inference_model(model, 'demo/demo.JPEG') +>>> print(predict_deploy['pred_class']) +sea snake +>>> print(predict_deploy['pred_score']) +0.4539395272731781 ``` -`${CFG_PATH}` is the config file path, `${SRC_CKPT_PATH}` is the source chenpoint file path, `${TARGET_CKPT_PATH}` is the target deploy weight file path. +**Use the model** -For example: +```python +>>> import torch +>>> from mmcls.apis import init_model +>>> +>>> model = init_model('configs/mobileone/mobileone-s0_8xb32_in1k.py', 'https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth') +>>> inputs = torch.rand(1, 3, 224, 224).to(model.data_preprocessor.device) +>>> # To get classification scores. +>>> out = model(inputs) +>>> print(out.shape) +torch.Size([1, 1000]) +>>> # To extract features. +>>> outs = model.extract_feat(inputs) +>>> print(outs[0].shape) +torch.Size([1, 768]) +>>> +>>> # switch to deploy mode +>>> model.backbone.switch_to_deploy() +>>> out_deploy = model(inputs) +>>> print(out.shape) +torch.Size([1, 1000]) +>>> assert torch.allclose(out, out_deploy) # pass without error +``` + +**Train/Test Command** + +Place the ImageNet dataset to the `data/imagenet/` directory, or prepare datasets according to the [docs](https://mmclassification.readthedocs.io/en/1.x/user_guides/dataset_prepare.html#prepare-dataset). + +Train: ```shell -python ./tools/convert_models/reparameterize_model.py ./configs/mobileone/mobileone-s0_8xb128_in1k.py https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_3rdparty_in1k_20220811-db5ce29b.pth ./mobileone_s0_deploy.pth +python tools/train.py configs/mobileone/mobileone-s0_8xb32_in1k.py ``` -To use reparameterized weights, the config file must switch to **the deploy config files**. +Download Checkpoint: -```bash -python tools/test.py ${Deploy_CFG} ${Deploy_Checkpoint} --metrics accuracy +```shell +wget https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth ``` -For example of using the reparameterized weights above: +Test use unfused model: ```shell -python ./tools/test.py ./configs/mobileone/deploy/mobileone-s0_deploy_8xb128_in1k.py mobileone_s0_deploy.pth --metrics accuracy +python tools/test.py configs/mobileone/mobileone-s0_8xb32_in1k.py mobileone-s0_8xb32_in1k_20221110-0bc94952.pth ``` -### In the code +Reparameterize checkpoint: -Use the API `switch_to_deploy` of `MobileOne` backbone to to switch to the deploy mode. Usually called like `backbone.switch_to_deploy()` or `classificer.backbone.switch_to_deploy()`. +```shell +python ./tools/convert_models/reparameterize_model.py ./configs/mobileone/mobileone-s0_8xb32_in1k.py mobileone-s0_8xb32_in1k_20221110-0bc94952.pth mobileone_s0_deploy.pth +``` -For Backbones: +Test use fused model: -```python -from mmcls.models import build_backbone -import torch +```shell +python tools/test.py configs/mobileone/deploy/mobileone-s0_deploy_8xb32_in1k.py mobileone_s0_deploy.pth +``` + + -x = torch.randn( (1, 3, 224, 224) ) -backbone_cfg=dict(type='MobileOne', arch='s0') -backbone = build_backbone(backbone_cfg) -backbone.init_weights() -backbone.eval() -outs_ori = backbone(x) +### Reparameterize Tool -backbone.switch_to_deploy() -outs_dep = backbone(x) +Use provided tool to reparameterize the given model and save the checkpoint: -for out1, out2 in zip(outs_ori, outs_dep): - assert torch.allclose(out1, out2) +```bash +python tools/convert_models/reparameterize_model.py ${CFG_PATH} ${SRC_CKPT_PATH} ${TARGET_CKPT_PATH} ``` -For ImageClassifiers: +`${CFG_PATH}` is the config file path, `${SRC_CKPT_PATH}` is the source chenpoint file path, `${TARGET_CKPT_PATH}` is the target deploy weight file path. -```python -from mmcls.models import build_classifier -import torch -import numpy as np - -cfg = dict( - type='ImageClassifier', - backbone=dict( - type='MobileOne', - arch='s0', - out_indices=(3, ), - ), - neck=dict(type='GlobalAveragePooling'), - head=dict( - type='LinearClsHead', - num_classes=1000, - in_channels=1024, - loss=dict(type='CrossEntropyLoss', loss_weight=1.0), - topk=(1, 5), - )) - -x = torch.randn( (1, 3, 224, 224) ) -classifier = build_classifier(cfg) -classifier.init_weights() -classifier.eval() -y_ori = classifier(x, return_loss=False) - -classifier.backbone.switch_to_deploy() -y_dep = classifier(x, return_loss=False) - -for y1, y2 in zip(y_ori, y_dep): - assert np.allclose(y1, y2) +For example: + +```shell +wget https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth +python ./tools/convert_models/reparameterize_model.py ./configs/mobileone/mobileone-s0_8xb32_in1k.py mobileone-s0_8xb32_in1k_20221110-0bc94952.pth mobileone_s0_deploy.pth ``` +To use reparameterized weights, the config file must switch to [**the deploy config files**](./deploy/). + +```bash +python tools/test.py ${Deploy_CFG} ${Deploy_Checkpoint} +``` + +For example of using the reparameterized weights above: + +```shell +python ./tools/test.py ./configs/mobileone/deploy/mobileone-s0_deploy_8xb32_in1k.py mobileone_s0_deploy.pth +``` + +For more configurable parameters, please refer to the [API](https://mmclassification.readthedocs.io/en/1.x/api/generated/mmcls.models.backbones.MobileOne.html#mmcls.models.backbones.MobileOne). + +## Results and models + +### ImageNet-1k + +| Model | Params(M) | Flops(G) | Top-1 (%) | Top-5 (%) | Config | Download | +| :----------: | :-----------------------------: | :----------------------------: | :-------: | :-------: | :---------------------------------------------------: | :------------------------------------------------------: | +| MobileOne-s0 | 5.29（train) \| 2.08 (deploy) | 1.09 (train) \| 0.28 (deploy) | 71.34 | 89.87 | [config (train)](./mobileone-s0_8xb32_in1k.py) \| [config (deploy)](./deploy/mobileone-s0_deploy_8xb32_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth) \| [log](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.json) | +| MobileOne-s1 | 4.83 (train) \| 4.76 (deploy) | 0.86 (train) \| 0.84 (deploy) | 75.72 | 92.54 | [config (train)](./mobileone-s1_8xb32_in1k.py) \| [config (deploy)](./deploy/mobileone-s1_deploy_8xb32_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s1_8xb32_in1k_20221110-ceeef467.pth) \| [log](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s1_8xb32_in1k_20221110-ceeef467.json) | +| MobileOne-s2 | 7.88 (train) \| 7.88 (deploy) | 1.34 (train) \| 1.31 (deploy) | 77.37 | 93.34 | [config (train)](./mobileone-s2_8xb32_in1k.py) \|[config (deploy)](./deploy/mobileone-s2_deploy_8xb32_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s2_8xb32_in1k_20221110-9c7ecb97.pth) \| [log](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s2_8xb32_in1k_20221110-9c7ecb97.json) | +| MobileOne-s3 | 10.17 (train) \| 10.08 (deploy) | 1.95 (train) \| 1.91 (deploy) | 78.06 | 93.83 | [config (train)](./mobileone-s3_8xb32_in1k.py) \|[config (deploy)](./deploy/mobileone-s3_deploy_8xb32_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s3_8xb32_in1k_20221110-c95eb3bf.pth) \| [log](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s3_8xb32_in1k_20221110-c95eb3bf.pth) | +| MobileOne-s4 | 14.95 (train) \| 14.84 (deploy) | 3.05 (train) \| 3.00 (deploy) | 79.69 | 94.46 | [config (train)](./mobileone-s4_8xb32_in1k.py) \|[config (deploy)](./deploy/mobileone-s4_deploy_8xb32_in1k.py) | [model](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s4_8xb32_in1k_20221110-28d888cb.pth) \| [log](https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s4_8xb32_in1k_20221110-28d888cb.pth) | + ## Citation ```bibtex diff --git a/configs/mobileone/deploy/mobileone-s0_deploy_8xb128_in1k.py b/configs/mobileone/deploy/mobileone-s0_deploy_8xb128_in1k.py deleted file mode 100644 index 8902483cfd2..00000000000 --- a/configs/mobileone/deploy/mobileone-s0_deploy_8xb128_in1k.py +++ /dev/null @@ -1,3 +0,0 @@ -_base_ = ['../mobileone-s0_8xb128_in1k.py'] - -model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s0_deploy_8xb32_in1k.py b/configs/mobileone/deploy/mobileone-s0_deploy_8xb32_in1k.py new file mode 100644 index 00000000000..145f3f4ec90 --- /dev/null +++ b/configs/mobileone/deploy/mobileone-s0_deploy_8xb32_in1k.py @@ -0,0 +1,3 @@ +_base_ = ['../mobileone-s0_8xb32_in1k.py'] + +model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s1_deploy_8xb128_in1k.py b/configs/mobileone/deploy/mobileone-s1_deploy_8xb128_in1k.py deleted file mode 100644 index 7bcf3211a13..00000000000 --- a/configs/mobileone/deploy/mobileone-s1_deploy_8xb128_in1k.py +++ /dev/null @@ -1,3 +0,0 @@ -_base_ = ['../mobileone-s1_8xb128_in1k.py'] - -model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s1_deploy_8xb32_in1k.py b/configs/mobileone/deploy/mobileone-s1_deploy_8xb32_in1k.py new file mode 100644 index 00000000000..8602c31ce6c --- /dev/null +++ b/configs/mobileone/deploy/mobileone-s1_deploy_8xb32_in1k.py @@ -0,0 +1,3 @@ +_base_ = ['../mobileone-s1_8xb32_in1k.py'] + +model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s2_deploy_8xb128_in1k.py b/configs/mobileone/deploy/mobileone-s2_deploy_8xb128_in1k.py deleted file mode 100644 index 5d64d51937c..00000000000 --- a/configs/mobileone/deploy/mobileone-s2_deploy_8xb128_in1k.py +++ /dev/null @@ -1,3 +0,0 @@ -_base_ = ['../mobileone-s2_8xb128_in1k.py'] - -model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s2_deploy_8xb32_in1k.py b/configs/mobileone/deploy/mobileone-s2_deploy_8xb32_in1k.py new file mode 100644 index 00000000000..97aaddd0740 --- /dev/null +++ b/configs/mobileone/deploy/mobileone-s2_deploy_8xb32_in1k.py @@ -0,0 +1,3 @@ +_base_ = ['../mobileone-s2_8xb32_in1k.py'] + +model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s3_deploy_8xb128_in1k.py b/configs/mobileone/deploy/mobileone-s3_deploy_8xb128_in1k.py deleted file mode 100644 index 8c710f7836a..00000000000 --- a/configs/mobileone/deploy/mobileone-s3_deploy_8xb128_in1k.py +++ /dev/null @@ -1,3 +0,0 @@ -_base_ = ['../mobileone-s3_8xb128_in1k.py'] - -model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s3_deploy_8xb32_in1k.py b/configs/mobileone/deploy/mobileone-s3_deploy_8xb32_in1k.py new file mode 100644 index 00000000000..0d335a7ba93 --- /dev/null +++ b/configs/mobileone/deploy/mobileone-s3_deploy_8xb32_in1k.py @@ -0,0 +1,3 @@ +_base_ = ['../mobileone-s3_8xb32_in1k.py'] + +model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s4_deploy_8xb128_in1k.py b/configs/mobileone/deploy/mobileone-s4_deploy_8xb128_in1k.py deleted file mode 100644 index 6ca4d18e7d2..00000000000 --- a/configs/mobileone/deploy/mobileone-s4_deploy_8xb128_in1k.py +++ /dev/null @@ -1,3 +0,0 @@ -_base_ = ['../mobileone-s4_8xb128_in1k.py'] - -model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/deploy/mobileone-s4_deploy_8xb32_in1k.py b/configs/mobileone/deploy/mobileone-s4_deploy_8xb32_in1k.py new file mode 100644 index 00000000000..b82f5a9ac7e --- /dev/null +++ b/configs/mobileone/deploy/mobileone-s4_deploy_8xb32_in1k.py @@ -0,0 +1,3 @@ +_base_ = ['../mobileone-s4_8xb32_in1k.py'] + +model = dict(backbone=dict(deploy=True)) diff --git a/configs/mobileone/metafile.yml b/configs/mobileone/metafile.yml index 04eaceff2ab..2a480dcdcda 100644 --- a/configs/mobileone/metafile.yml +++ b/configs/mobileone/metafile.yml @@ -16,83 +16,68 @@ Collections: Version: v1.0.0rc1 Models: - - Name: mobileone-s0_3rdparty_8xb128_in1k + - Name: mobileone-s0_8xb32_in1k In Collection: MobileOne - Config: configs/mobileone/mobileone-s0_8xb128_in1k.py + Config: configs/mobileone/mobileone-s0_8xb32_in1k.py Metadata: - FLOPs: 1091227648 # 1.09G - Parameters: 5293272 # 5.29M + FLOPs: 274136576 # 0.27G + Parameters: 2078504 # 2.08M Results: - Dataset: ImageNet-1k Task: Image Classification Metrics: - Top 1 Accuracy: 71.36 + Top 1 Accuracy: 71.34 Top 5 Accuracy: 89.87 - Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_3rdparty_in1k_20220915-007ae971.pth - Converted From: - Weights: https://docs-assets.developer.apple.com/ml-research/datasets/mobileone/mobileone_s0_unfused.pth.tar - Code: https://github.com/apple/ml-mobileone - - Name: mobileone-s1_3rdparty_8xb128_in1k + Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s0_8xb32_in1k_20221110-0bc94952.pth + - Name: mobileone-s1_8xb32_in1k In Collection: MobileOne - Config: configs/mobileone/mobileone-s1_8xb128_in1k.py + Config: configs/mobileone/mobileone-s1_8xb32_in1k.py Metadata: - FLOPs: 863491328 # 8.6G - Parameters: 4825192 # 4.82M + FLOPs: 823839744 # 8.6G + Parameters: 4764840 # 4.82M Results: - Dataset: ImageNet-1k Task: Image Classification Metrics: - Top 1 Accuracy: 75.76 - Top 5 Accuracy: 92.77 - Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s1_3rdparty_in1k_20220915-473c8469.pth - Converted From: - Weights: https://docs-assets.developer.apple.com/ml-research/datasets/mobileone/mobileone_s1_unfused.pth.tar - Code: https://github.com/apple/ml-mobileone - - Name: mobileone-s2_3rdparty_8xb128_in1k + Top 1 Accuracy: 75.72 + Top 5 Accuracy: 92.54 + Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s1_8xb32_in1k_20221110-ceeef467.pth + - Name: mobileone-s2_8xb32_in1k In Collection: MobileOne - Config: configs/mobileone/mobileone-s2_8xb128_in1k.py + Config: configs/mobileone/mobileone-s2_8xb32_in1k.py Metadata: - FLOPs: 1344083328 - Parameters: 7884648 + FLOPs: 1296478848 + Parameters: 7808168 Results: - Dataset: ImageNet-1k Task: Image Classification Metrics: - Top 1 Accuracy: 77.39 - Top 5 Accuracy: 93.63 - Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s2_3rdparty_in1k_20220915-ed2e4c30.pth - Converted From: - Weights: https://docs-assets.developer.apple.com/ml-research/datasets/mobileone/mobileone_s2_unfused.pth.tar - Code: https://github.com/apple/ml-mobileone - - Name: mobileone-s3_3rdparty_8xb128_in1k + Top 1 Accuracy: 77.37 + Top 5 Accuracy: 93.34 + Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s2_8xb32_in1k_20221110-9c7ecb97.pth + - Name: mobileone-s3_8xb32_in1k In Collection: MobileOne - Config: configs/mobileone/mobileone-s3_8xb128_in1k.py + Config: configs/mobileone/mobileone-s3_8xb32_in1k.py Metadata: - FLOPs: 1951043584 - Parameters: 10170600 + FLOPs: 1893842944 + Parameters: 10078312 Results: - Dataset: ImageNet-1k Task: Image Classification Metrics: - Top 1 Accuracy: 77.93 - Top 5 Accuracy: 93.89 - Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s3_3rdparty_in1k_20220915-84d6a02c.pth - Converted From: - Weights: https://docs-assets.developer.apple.com/ml-research/datasets/mobileone/mobileone_s3_unfused.pth.tar - Code: https://github.com/apple/ml-mobileone - - Name: mobileone-s4_3rdparty_8xb128_in1k + Top 1 Accuracy: 78.06 + Top 5 Accuracy: 93.83 + Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s3_8xb32_in1k_20221110-c95eb3bf.pth + - Name: mobileone-s4_8xb32_in1k In Collection: MobileOne - Config: configs/mobileone/mobileone-s4_8xb128_in1k.py + Config: configs/mobileone/mobileone-s4_8xb32_in1k.py Metadata: - FLOPs: 3052580688 - Parameters: 14951248 + FLOPs: 2979222528 + Parameters: 14838352 Results: - Dataset: ImageNet-1k Task: Image Classification Metrics: - Top 1 Accuracy: 79.30 - Top 5 Accuracy: 94.37 - Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s4_3rdparty_in1k_20220915-ce9509ee.pth - Converted From: - Weights: https://docs-assets.developer.apple.com/ml-research/datasets/mobileone/mobileone_s4_unfused.pth.tar - Code: https://github.com/apple/ml-mobileone + Top 1 Accuracy: 79.69 + Top 5 Accuracy: 94.46 + Weights: https://download.openmmlab.com/mmclassification/v0/mobileone/mobileone-s4_8xb32_in1k_20221110-28d888cb.pth diff --git a/configs/mobileone/mobileone-s0_8xb32_in1k.py b/configs/mobileone/mobileone-s0_8xb32_in1k.py new file mode 100644 index 00000000000..be56b86c3ce --- /dev/null +++ b/configs/mobileone/mobileone-s0_8xb32_in1k.py @@ -0,0 +1,20 @@ +_base_ = [ + '../_base_/models/mobileone/mobileone_s0.py', + '../_base_/datasets/imagenet_bs32_pil_resize.py', + '../_base_/schedules/imagenet_bs256_coslr_coswd_300e.py', + '../_base_/default_runtime.py' +] + +# schedule settings +optim_wrapper = dict(paramwise_cfg=dict(norm_decay_mult=0.)) + +val_dataloader = dict(batch_size=256) +test_dataloader = dict(batch_size=256) + +custom_hooks = [ + dict( + type='EMAHook', + momentum=5e-4, + priority='ABOVE_NORMAL', + update_buffers=True) +] diff --git a/configs/mobileone/mobileone-s1_8xb128_in1k.py b/configs/mobileone/mobileone-s1_8xb128_in1k.py deleted file mode 100644 index b14c7c17d1b..00000000000 --- a/configs/mobileone/mobileone-s1_8xb128_in1k.py +++ /dev/null @@ -1,15 +0,0 @@ -_base_ = [ - '../_base_/models/mobileone/mobileone_s1.py', - '../_base_/datasets/imagenet_bs32_pil_resize.py', - '../_base_/schedules/imagenet_bs256_coslr.py', - '../_base_/default_runtime.py' -] - -# dataset settings -train_dataloader = dict(batch_size=128) -val_dataloader = dict(batch_size=128) -test_dataloader = dict(batch_size=128) - -# NOTE: `auto_scale_lr` is for automatically scaling LR, -# based on the actual training batch size. -auto_scale_lr = dict(base_batch_size=1024) diff --git a/configs/mobileone/mobileone-s1_8xb32_in1k.py b/configs/mobileone/mobileone-s1_8xb32_in1k.py new file mode 100644 index 00000000000..52c8442e82d --- /dev/null +++ b/configs/mobileone/mobileone-s1_8xb32_in1k.py @@ -0,0 +1,60 @@ +_base_ = [ + '../_base_/models/mobileone/mobileone_s1.py', + '../_base_/datasets/imagenet_bs32_pil_resize.py', + '../_base_/schedules/imagenet_bs256_coslr_coswd_300e.py', + '../_base_/default_runtime.py' +] + +# schedule settings +optim_wrapper = dict(paramwise_cfg=dict(norm_decay_mult=0.)) + +val_dataloader = dict(batch_size=256) +test_dataloader = dict(batch_size=256) + +bgr_mean = _base_.data_preprocessor['mean'][::-1] +base_train_pipeline = [ + dict(type='LoadImageFromFile'), + dict(type='RandomResizedCrop', scale=224, backend='pillow'), + dict(type='RandomFlip', prob=0.5, direction='horizontal'), + dict( + type='RandAugment', + policies='timm_increasing', + num_policies=2, + total_level=10, + magnitude_level=7, + magnitude_std=0.5, + hparams=dict(pad_val=[round(x) for x in bgr_mean])), + dict(type='PackClsInputs') +] + +import copy # noqa: E402 + +# modify start epoch's RandomResizedCrop.scale to 160 +train_pipeline_1e = copy.deepcopy(base_train_pipeline) +train_pipeline_1e[1]['scale'] = 160 +train_pipeline_1e[3]['magnitude_level'] *= 0.1 +_base_.train_dataloader.dataset.pipeline = train_pipeline_1e + +# modify 37 epoch's RandomResizedCrop.scale to 192 +train_pipeline_37e = copy.deepcopy(base_train_pipeline) +train_pipeline_37e[1]['scale'] = 192 +train_pipeline_1e[3]['magnitude_level'] *= 0.2 + +# modify 112 epoch's RandomResizedCrop.scale to 224 +train_pipeline_112e = copy.deepcopy(base_train_pipeline) +train_pipeline_112e[1]['scale'] = 224 +train_pipeline_1e[3]['magnitude_level'] *= 0.3 + +custom_hooks = [ + dict( + type='SwitchRecipeHook', + schedule=[ + dict(action_epoch=37, pipeline=train_pipeline_37e), + dict(action_epoch=112, pipeline=train_pipeline_112e), + ]), + dict( + type='EMAHook', + momentum=5e-4, + priority='ABOVE_NORMAL', + update_buffers=True) +] diff --git a/configs/mobileone/mobileone-s2_8xb128_in1k.py b/configs/mobileone/mobileone-s2_8xb128_in1k.py deleted file mode 100644 index dca0d4d3e58..00000000000 --- a/configs/mobileone/mobileone-s2_8xb128_in1k.py +++ /dev/null @@ -1,15 +0,0 @@ -_base_ = [ - '../_base_/models/mobileone/mobileone_s2.py', - '../_base_/datasets/imagenet_bs32_pil_resize.py', - '../_base_/schedules/imagenet_bs256_coslr.py', - '../_base_/default_runtime.py' -] - -# dataset settings -train_dataloader = dict(batch_size=128) -val_dataloader = dict(batch_size=128) -test_dataloader = dict(batch_size=128) - -# NOTE: `auto_scale_lr` is for automatically scaling LR, -# based on the actual training batch size. -auto_scale_lr = dict(base_batch_size=1024) diff --git a/configs/mobileone/mobileone-s2_8xb32_in1k.py b/configs/mobileone/mobileone-s2_8xb32_in1k.py new file mode 100644 index 00000000000..547ae9952d1 --- /dev/null +++ b/configs/mobileone/mobileone-s2_8xb32_in1k.py @@ -0,0 +1,65 @@ +_base_ = [ + '../_base_/models/mobileone/mobileone_s2.py', + '../_base_/datasets/imagenet_bs32_pil_resize.py', + '../_base_/schedules/imagenet_bs256_coslr_coswd_300e.py', + '../_base_/default_runtime.py' +] + +# schedule settings +optim_wrapper = dict(paramwise_cfg=dict(norm_decay_mult=0.)) + +val_dataloader = dict(batch_size=256) +test_dataloader = dict(batch_size=256) + +import copy # noqa: E402 + +bgr_mean = _base_.data_preprocessor['mean'][::-1] +base_train_pipeline = [ + dict(type='LoadImageFromFile'), + dict(type='RandomResizedCrop', scale=224, backend='pillow'), + dict(type='RandomFlip', prob=0.5, direction='horizontal'), + dict( + type='RandAugment', + policies='timm_increasing', + num_policies=2, + total_level=10, + magnitude_level=7, + magnitude_std=0.5, + hparams=dict(pad_val=[round(x) for x in bgr_mean])), + dict(type='PackClsInputs') +] + +# modify start epoch RandomResizedCrop.scale to 160 +# and RA.magnitude_level * 0.3 +train_pipeline_1e = copy.deepcopy(base_train_pipeline) +train_pipeline_1e[1]['scale'] = 160 +train_pipeline_1e[3]['magnitude_level'] *= 0.3 +_base_.train_dataloader.dataset.pipeline = train_pipeline_1e + +import copy # noqa: E402 + +# modify 137 epoch's RandomResizedCrop.scale to 192 +# and RA.magnitude_level * 0.7 +train_pipeline_37e = copy.deepcopy(base_train_pipeline) +train_pipeline_37e[1]['scale'] = 192 +train_pipeline_37e[3]['magnitude_level'] *= 0.7 + +# modify 112 epoch's RandomResizedCrop.scale to 224 +# and RA.magnitude_level * 1.0 +train_pipeline_112e = copy.deepcopy(base_train_pipeline) +train_pipeline_112e[1]['scale'] = 224 +train_pipeline_112e[3]['magnitude_level'] *= 1.0 + +custom_hooks = [ + dict( + type='SwitchRecipeHook', + schedule=[ + dict(action_epoch=37, pipeline=train_pipeline_37e), + dict(action_epoch=112, pipeline=train_pipeline_112e), + ]), + dict( + type='EMAHook', + momentum=5e-4, + priority='ABOVE_NORMAL', + update_buffers=True) +] diff --git a/configs/mobileone/mobileone-s3_8xb128_in1k.py b/configs/mobileone/mobileone-s3_8xb128_in1k.py deleted file mode 100644 index 89343d5d861..00000000000 --- a/configs/mobileone/mobileone-s3_8xb128_in1k.py +++ /dev/null @@ -1,15 +0,0 @@ -_base_ = [ - '../_base_/models/mobileone/mobileone_s3.py', - '../_base_/datasets/imagenet_bs64_pil_resize.py', - '../_base_/schedules/imagenet_bs256_coslr.py', - '../_base_/default_runtime.py' -] - -# dataset settings -train_dataloader = dict(batch_size=128) -val_dataloader = dict(batch_size=128) -test_dataloader = dict(batch_size=128) - -# NOTE: `auto_scale_lr` is for automatically scaling LR, -# based on the actual training batch size. -auto_scale_lr = dict(base_batch_size=1024) diff --git a/configs/mobileone/mobileone-s3_8xb32_in1k.py b/configs/mobileone/mobileone-s3_8xb32_in1k.py new file mode 100644 index 00000000000..b0ef416469f --- /dev/null +++ b/configs/mobileone/mobileone-s3_8xb32_in1k.py @@ -0,0 +1,65 @@ +_base_ = [ + '../_base_/models/mobileone/mobileone_s3.py', + '../_base_/datasets/imagenet_bs32_pil_resize.py', + '../_base_/schedules/imagenet_bs256_coslr_coswd_300e.py', + '../_base_/default_runtime.py' +] + +# schedule settings +optim_wrapper = dict(paramwise_cfg=dict(norm_decay_mult=0.)) + +val_dataloader = dict(batch_size=256) +test_dataloader = dict(batch_size=256) + +import copy # noqa: E402 + +bgr_mean = _base_.data_preprocessor['mean'][::-1] +base_train_pipeline = [ + dict(type='LoadImageFromFile'), + dict(type='RandomResizedCrop', scale=224, backend='pillow'), + dict(type='RandomFlip', prob=0.5, direction='horizontal'), + dict( + type='RandAugment', + policies='timm_increasing', + num_policies=2, + total_level=10, + magnitude_level=7, + magnitude_std=0.5, + hparams=dict(pad_val=[round(x) for x in bgr_mean])), + dict(type='PackClsInputs') +] + +# modify start epoch RandomResizedCrop.scale to 160 +# and RA.magnitude_level * 0.3 +train_pipeline_1e = copy.deepcopy(base_train_pipeline) +train_pipeline_1e[1]['scale'] = 160 +train_pipeline_1e[3]['magnitude_level'] *= 0.3 +_base_.train_dataloader.dataset.pipeline = train_pipeline_1e + +import copy # noqa: E402 + +# modify 137 epoch's RandomResizedCrop.scale to 192 +# and RA.magnitude_level * 0.7 +train_pipeline_37e = copy.deepcopy(base_train_pipeline) +train_pipeline_37e[1]['scale'] = 192 +train_pipeline_37e[3]['magnitude_level'] *= 0.7 + +# modify 112 epoch's RandomResizedCrop.scale to 224 +# and RA.magnitude_level * 1.0 +train_pipeline_112e = copy.deepcopy(base_train_pipeline) +train_pipeline_112e[1]['scale'] = 224 +train_pipeline_112e[3]['magnitude_level'] *= 1.0 + +custom_hooks = [ + dict( + type='SwitchRecipeHook', + schedule=[ + dict(action_epoch=37, pipeline=train_pipeline_37e), + dict(action_epoch=112, pipeline=train_pipeline_112e), + ]), + dict( + type='EMAHook', + momentum=5e-4, + priority='ABOVE_NORMAL', + update_buffers=True) +] diff --git a/configs/mobileone/mobileone-s4_8xb128_in1k.py b/configs/mobileone/mobileone-s4_8xb128_in1k.py deleted file mode 100644 index 1984ef351de..00000000000 --- a/configs/mobileone/mobileone-s4_8xb128_in1k.py +++ /dev/null @@ -1,15 +0,0 @@ -_base_ = [ - '../_base_/models/mobileone/mobileone_s4.py', - '../_base_/datasets/imagenet_bs64_pil_resize.py', - '../_base_/schedules/imagenet_bs256_coslr.py', - '../_base_/default_runtime.py' -] - -# dataset settings -train_dataloader = dict(batch_size=128) -val_dataloader = dict(batch_size=128) -test_dataloader = dict(batch_size=128) - -# NOTE: `auto_scale_lr` is for automatically scaling LR, -# based on the actual training batch size. -auto_scale_lr = dict(base_batch_size=1024) diff --git a/configs/mobileone/mobileone-s4_8xb32_in1k.py b/configs/mobileone/mobileone-s4_8xb32_in1k.py new file mode 100644 index 00000000000..8c31f2400a8 --- /dev/null +++ b/configs/mobileone/mobileone-s4_8xb32_in1k.py @@ -0,0 +1,63 @@ +_base_ = [ + '../_base_/models/mobileone/mobileone_s4.py', + '../_base_/datasets/imagenet_bs32_pil_resize.py', + '../_base_/schedules/imagenet_bs256_coslr_coswd_300e.py', + '../_base_/default_runtime.py' +] + +# schedule settings +optim_wrapper = dict(paramwise_cfg=dict(norm_decay_mult=0.)) + +val_dataloader = dict(batch_size=256) +test_dataloader = dict(batch_size=256) + +bgr_mean = _base_.data_preprocessor['mean'][::-1] +base_train_pipeline = [ + dict(type='LoadImageFromFile'), + dict(type='RandomResizedCrop', scale=224, backend='pillow'), + dict(type='RandomFlip', prob=0.5, direction='horizontal'), + dict( + type='RandAugment', + policies='timm_increasing', + num_policies=2, + total_level=10, + magnitude_level=7, + magnitude_std=0.5, + hparams=dict(pad_val=[round(x) for x in bgr_mean])), + dict(type='PackClsInputs') +] + +import copy # noqa: E402 + +# modify start epoch RandomResizedCrop.scale to 160 +# and RA.magnitude_level * 0.3 +train_pipeline_1e = copy.deepcopy(base_train_pipeline) +train_pipeline_1e[1]['scale'] = 160 +train_pipeline_1e[3]['magnitude_level'] *= 0.3 +_base_.train_dataloader.dataset.pipeline = train_pipeline_1e + +# modify 137 epoch's RandomResizedCrop.scale to 192 +# and RA.magnitude_level * 0.7 +train_pipeline_37e = copy.deepcopy(base_train_pipeline) +train_pipeline_37e[1]['scale'] = 192 +train_pipeline_37e[3]['magnitude_level'] *= 0.7 + +# modify 112 epoch's RandomResizedCrop.scale to 224 +# and RA.magnitude_level * 1.0 +train_pipeline_112e = copy.deepcopy(base_train_pipeline) +train_pipeline_112e[1]['scale'] = 224 +train_pipeline_112e[3]['magnitude_level'] *= 1.0 + +custom_hooks = [ + dict( + type='SwitchRecipeHook', + schedule=[ + dict(action_epoch=37, pipeline=train_pipeline_37e), + dict(action_epoch=112, pipeline=train_pipeline_112e), + ]), + dict( + type='EMAHook', + momentum=5e-4, + priority='ABOVE_NORMAL', + update_buffers=True) +] diff --git a/tools/visualizations/vis_scheduler.py b/tools/visualizations/vis_scheduler.py index 87d076dcba6..5e08af39bdc 100644 --- a/tools/visualizations/vis_scheduler.py +++ b/tools/visualizations/vis_scheduler.py @@ -41,6 +41,7 @@ def __init__(self, by_epoch): self.by_epoch = by_epoch self.lr_list = [] self.momentum_list = [] + self.wd_list = [] self.task_id = 0 self.progress = Progress(BarColumn(), MofNCompleteColumn(), TextColumn('{task.description}')) @@ -66,6 +67,8 @@ def after_train_iter(self, runner, batch_idx, data_batch, outputs): self.lr_list.append(runner.optim_wrapper.get_lr()['lr'][0]) self.momentum_list.append( runner.optim_wrapper.get_momentum()['momentum'][0]) + self.wd_list.append( + runner.optim_wrapper.param_groups[0]['weight_decay']) def after_train(self, runner): self.progress.stop() @@ -80,9 +83,9 @@ def parse_args(): '--parameter', type=str, default='lr', - choices=['lr', 'momentum'], + choices=['lr', 'momentum', 'wd'], help='The parameter to visualize its change curve, choose from' - '"lr" and "momentum". Defaults to "lr".') + '"lr", "wd" and "momentum". Defaults to "lr".') parser.add_argument( '-d', '--dataset-size', @@ -192,7 +195,12 @@ def simulate_train(data_loader, cfg, by_epoch): runner.train() - return param_record_hook.lr_list, param_record_hook.momentum_list + param_dict = dict( + lr=param_record_hook.lr_list, + momentum=param_record_hook.momentum_list, + wd=param_record_hook.wd_list) + + return param_dict def main(): @@ -250,13 +258,15 @@ class FakeDataloader(list): rich.print(dataset_info + '\n') # simulation training process - lr_list, momentum_list = simulate_train(data_loader, cfg, by_epoch) + param_dict = simulate_train(data_loader, cfg, by_epoch) + param_list = param_dict[args.parameter] + if args.parameter == 'lr': - param_list = lr_list + param_name = 'Learning Rate' + elif args.parameter == 'momentum': + param_name = 'Momentum' else: - param_list = momentum_list - - param_name = 'Learning Rate' if args.parameter == 'lr' else 'Momentum' + param_name = 'Weight Decay' plot_curve(param_list, args, param_name, len(data_loader), by_epoch) if args.save_path: