Skip to content

Commit

Permalink
update release for det and seg
Browse files Browse the repository at this point in the history
  • Loading branch information
@Lupin1998
Lupin1998 committed Feb 9, 2023
1 parent 23d17b6 commit 5a72c32
Show file tree
Hide file tree
Showing 4 changed files with 39 additions and 32 deletions.
8 changes: 4 additions & 4 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,10 +38,10 @@ We propose **MogaNet**, a new family of efficient ConvNets, to pursue informativ

We plan to release implementations of MogaNet in a few months. Please watch us for the latest release. Currently, this repo is reimplemented according to our official implementations in [OpenMixup](https://github.com/Westlake-AI/openmixup/), and we are working on cleaning up experimental results and code implementations.

- [x] **ImageNet-1K** Training and Validation Code [[here](#image-classification)]
- [x] Downstream Transfer to **Object Detection and Instance Segmentation on COCO** [[here](detection/)]
- [x] Downstream Transfer to **Semantic Segmentation on ADE20K** [[here](segmentation/)]
- [x] Downstream Transfer to **2D Human Pose Estimation on COCO** [[here](pose_estimation/)]
- [x] **ImageNet-1K** Training and Validation Code [[here](#image-classification)] [[models](https://github.com/Westlake-AI/MogaNet/releases/tag/moganet-in1k-weights)]
- [x] Downstream Transfer to **Object Detection and Instance Segmentation on COCO** [[code](detection/)] [[models](https://github.com/Westlake-AI/MogaNet/releases/tag/moganet-det-weights)]
- [x] Downstream Transfer to **Semantic Segmentation on ADE20K** [[code](segmentation/)] [[models](https://github.com/Westlake-AI/MogaNet/releases/tag/moganet-seg-weights)]
- [x] Downstream Transfer to **2D Human Pose Estimation on COCO** [[code](pose_estimation/)]
- [x] Image Classification on Google Colab and Notebook Demo [[here](demo.ipynb)]

## Image Classification
Expand Down
37 changes: 22 additions & 15 deletions detection/README.md
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# Applying MogaNet to Object Detection

This repo is a PyTorch implementation of applying **MogaNet** to object detaction and instance segmentation on COCO. The code is based on [MMDetection](https://github.com/open-mmlab/mmdetection/tree/v2.26.0).
This repo is a PyTorch implementation of applying **MogaNet** to object detaction and instance segmentation with [Mask R-CNN](https://arxiv.org/abs/1703.06870) and [RetinaNet](https://arxiv.org/abs/1708.02002) on [COCO](https://arxiv.org/abs/1405.0312). The code is based on [MMDetection](https://github.com/open-mmlab/mmdetection/tree/v2.26.0).
For more details, see [Efficient Multi-order Gated Aggregation Network](https://arxiv.org/abs/2211.03295) (arXiv 2022).

## Note
Expand Down Expand Up @@ -40,30 +40,37 @@ Note: Since we write [MogaNet backbone code](../models/moganet.py) of detection,

## Data preparation

Prepare COCO according to the guidelines in [MMDetection](https://github.com/open-mmlab/mmdetection/).
Download [COCO2017](https://cocodataset.org/#download) and prepare COCO experiments according to the guidelines in [MMDetection](https://github.com/open-mmlab/mmdetection/).

<p align="right">(<a href="#top">back to top</a>)</p>

## Results and models on COCO

| Method | Backbone | Pretrain | Params | FLOPs | Lr schd | Aug | box mAP | mask mAP | Config | Download |
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
| RetinaNet | MogaNet-XT | ImageNet-1K | 12.1M | 167.2G | 1x | No | 39.7 | - | [config](configs/mask_rcnn_moganet_xtiny_fpn_1x_coco.py) | log / model |
| RetinaNet | MogaNet-T | ImageNet-1K | 14.4M | 173.4G | 1x | No | 41.4 | - | [config](configs/mask_rcnn_moganet_tiny_fpn_1x_coco.py) | log / model |
| RetinaNet | MogaNet-S | ImageNet-1K | 35.1M | 253.0G | 1x | No | 45.8 | - | [config](configs/mask_rcnn_moganet_small_fpn_1x_coco.py) | log / model |
| RetinaNet | MogaNet-B | ImageNet-1K | 53.5M | 354.5G | 1x | No | 47.7 | - | [config](configs/mask_rcnn_moganet_base_fpn_1x_coco.py) | log / model |
| RetinaNet | MogaNet-L | ImageNet-1K | 92.4M | 476.8G | 1x | No | 48.7 | - | [config](configs/mask_rcnn_moganet_large_fpn_1x_coco.py) | log / model |
| Mask R-CNN | MogaNet-XT | ImageNet-1K | 22.8M | 185.4G | 1x | No | 40.7 | 37.6 | [config](configs/retinanet_moganet_xtiny_fpn_1x_coco.py) | log / model |
| Mask R-CNN | MogaNet-T | ImageNet-1K | 25.0M | 191.7G | 1x | No | 42.6 | 39.1 | [config](configs/retinanet_moganet_tiny_fpn_1x_coco.py) | log / model |
| Mask R-CNN | MogaNet-S | ImageNet-1K | 45.0M | 271.6G | 1x | No | 46.6 | 42.2 | [config](configs/retinanet_moganet_small_fpn_1x_coco.py) | log / model |
| Mask R-CNN | MogaNet-B | ImageNet-1K | 63.4M | 373.1G | 1x | No | 49.0 | 43.8 | [config](configs/retinanet_moganet_base_fpn_1x_coco.py) | log / model |
| Mask R-CNN | MogaNet-L | ImageNet-1K | 102.1M | 495.3G | 1x | No | 49.4 | 44.2 | [config](configs/retinanet_moganet_large_fpn_1x_coco.py) | log / model |

**Notes**: All the models can also be downloaded by [**Baidu Cloud**](https://pan.baidu.com/s/1d5MTTC66gegehmfZvCQRUA?pwd=z8mf) (z8mf) at `MogaNet/COCO_Detection`. The params (M) and FLOPs (G) are measured by [get_flops](get_flops.sh) with 1280 $\times$ 800 resolutions.
```bash
bash get_flops.sh /path/to/config --shape 1280 800
```

### MogaNet + RetinaNet

| Method | Backbone | Pretrain | Params | FLOPs | Lr schd | box mAP | Config | Download |
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
| RetinaNet | MogaNet-XT | ImageNet-1K | 12.1M | 167.2G | 1x | 39.7 | [config](configs/retinanet_moganet_xtiny_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_xtiny_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_xtiny_fpn_1x_coco.pth) |
| RetinaNet | MogaNet-T | ImageNet-1K | 14.4M | 173.4G | 1x | 41.4 | [config](configs/retinanet_moganet_tiny_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_tiny_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_tiny_fpn_1x_coco.pth) |
| RetinaNet | MogaNet-S | ImageNet-1K | 35.1M | 253.0G | 1x | 45.8 | [config](configs/retinanet_moganet_small_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_small_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_small_fpn_1x_coco.pth) |
| RetinaNet | MogaNet-B | ImageNet-1K | 53.5M | 354.5G | 1x | 47.7 | [config](configs/retinanet_moganet_base_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_base_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_base_fpn_1x_coco.pth) |
| RetinaNet | MogaNet-L | ImageNet-1K | 92.4M | 476.8G | 1x | 48.7 | [config](configs/retinanet_moganet_large_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_large_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/retinanet_moganet_large_fpn_1x_coco.pth) |

### MogaNet + Mask R-CNN

| Method | Backbone | Pretrain | Params | FLOPs | Lr schd | box mAP | mask mAP | Config | Download |
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
| Mask R-CNN | MogaNet-XT | ImageNet-1K | 22.8M | 185.4G | 1x | 40.7 | 37.6 | [config](configs/mask_rcnn_moganet_xtiny_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_xtiny_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_xtiny_fpn_1x_coco.pth) |
| Mask R-CNN | MogaNet-T | ImageNet-1K | 25.0M | 191.7G | 1x | 42.6 | 39.1 | [config](configs/mask_rcnn_moganet_tiny_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_tiny_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_tiny_fpn_1x_coco.pth) |
| Mask R-CNN | MogaNet-S | ImageNet-1K | 45.0M | 271.6G | 1x | 46.6 | 42.2 | [config](configs/mask_rcnn_moganet_small_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_small_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_small_fpn_1x_coco.pth) |
| Mask R-CNN | MogaNet-B | ImageNet-1K | 63.4M | 373.1G | 1x | 49.0 | 43.8 | [config](configs/mask_rcnn_moganet_base_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_base_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_base_fpn_1x_coco.pth) |
| Mask R-CNN | MogaNet-L | ImageNet-1K | 102.1M | 495.3G | 1x | 49.4 | 44.2 | [config](configs/mask_rcnn_moganet_large_fpn_1x_coco.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_large_fpn_1x_coco.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-det-weights/mask_rcnn_moganet_large_fpn_1x_coco.pth) |

## Training

We train the model on a single node with 8 GPUs (a batch size of 16) by default. Start training with the config as:
Expand Down
2 changes: 1 addition & 1 deletion pose_estimation/README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ Note: Since we write [MogaNet backbone code](../models/moganet.py) of detection,

## Data preparation

Prepare COCO according to the guidelines in [MMPose](https://github.com/open-mmlab/mmpose/).
Download [COCO2017](https://cocodataset.org/#download) and prepare COCO experiments according to the guidelines in [MMPose](https://github.com/open-mmlab/mmpose/).

<p align="right">(<a href="#top">back to top</a>)</p>

Expand Down
24 changes: 12 additions & 12 deletions segmentation/README.md
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# Applying MogaNet to Semantic Segmentation

This repo is a PyTorch implementation of applying **MogaNet** to semantic segmentation on ADE20K. The code is based on [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/tree/v0.29.1).
This repo is a PyTorch implementation of applying **MogaNet** to semantic segmentation with [Semantic FPN](https://arxiv.org/abs/1901.02446) and [UperNet](https://arxiv.org/abs/1505.04597) on [ADE20K](https://arxiv.org/abs/1608.05442). The code is based on [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/tree/v0.29.1).
For more details, see [Efficient Multi-order Gated Aggregation Network](https://arxiv.org/abs/2211.03295) (arXiv 2022).

## Note
Expand All @@ -26,7 +26,7 @@ Prepare ADE20K according to the [guidelines](https://github.com/open-mmlab/mmseg

## Results and models on ADE20K

**Notes**: All the models are evaluated at a single scale (SS), you can modify `test_pipeline` to evaluate the multi-scale performance (MS). The trained models can also be downloaded by [**Baidu Cloud**](https://pan.baidu.com/s/1d5MTTC66gegehmfZvCQRUA?pwd=z8mf) (z8mf) at `MogaNet/ADE20K_Segmentation`. The params (M) and FLOPs (G) are measured by [get_flops](get_flops.sh) with 2048 $\times$ 512 resolutions.
**Notes**: All the models are evaluated at a single scale (SS), you can modify `test_pipeline` in config files to evaluate the multi-scale performance (MS). The trained models can also be downloaded by [**Baidu Cloud**](https://pan.baidu.com/s/1d5MTTC66gegehmfZvCQRUA?pwd=z8mf) (z8mf) at `MogaNet/ADE20K_Segmentation`. The params (M) and FLOPs (G) are measured by [get_flops](get_flops.sh) with 2048 $\times$ 512 resolutions.
```bash
bash get_flops.sh /path/to/config --shape 2048 512
```
Expand All @@ -35,21 +35,21 @@ bash get_flops.sh /path/to/config --shape 2048 512

| Method | Backbone | Pretrain | Params | FLOPs | Iters | mIoU | mAcc | Config | Download |
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
| Semantic FPN | MogaNet-XT | ImageNet-1K | 6.9M | 101.4G | 80K | 40.3 | 52.4 | [config](configs/sem_fpn/moganet/fpn_moganet_xtiny_80k_ade20k.py) | log / model |
| Semantic FPN | MogaNet-T | ImageNet-1K | 9.1M | 107.8G | 80K | 43.1 | 55.4 | [config](configs/sem_fpn/moganet/fpn_moganet_tiny_80k_ade20k.py) | log / model |
| Semantic FPN | MogaNet-S | ImageNet-1K | 29.1M | 189.7G | 80K | 47.7 | 59.8 | [config](configs/sem_fpn/moganet/fpn_moganet_small_80k_ade20k.py) | log / model |
| Semantic FPN | MogaNet-B | ImageNet-1K | 47.5M | 293.6G | 80K | 49.3 | 61.6 | [config](configs/sem_fpn/moganet/fpn_moganet_base_80k_ade20k.py) | log / model |
| Semantic FPN | MogaNet-L | ImageNet-1K | 86.2M | 418.7G | 80K | 50.2 | 63.0 | [config](configs/sem_fpn/moganet/fpn_moganet_large_80k_ade20k.py) | log / model |
| Semantic FPN | MogaNet-XT | ImageNet-1K | 6.9M | 101.4G | 80K | 40.3 | 52.4 | [config](configs/sem_fpn/moganet/fpn_moganet_xtiny_80k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_xtiny_80k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_xtiny_80k_ade20k.pth) |
| Semantic FPN | MogaNet-T | ImageNet-1K | 9.1M | 107.8G | 80K | 43.1 | 55.4 | [config](configs/sem_fpn/moganet/fpn_moganet_tiny_80k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_tiny_80k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_tiny_80k_ade20k.pth) |
| Semantic FPN | MogaNet-S | ImageNet-1K | 29.1M | 189.7G | 80K | 47.7 | 59.8 | [config](configs/sem_fpn/moganet/fpn_moganet_small_80k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_small_80k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_small_80k_ade20k.pth) |
| Semantic FPN | MogaNet-B | ImageNet-1K | 47.5M | 293.6G | 80K | 49.3 | 61.6 | [config](configs/sem_fpn/moganet/fpn_moganet_base_80k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_base_80k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_base_80k_ade20k.pth) |
| Semantic FPN | MogaNet-L | ImageNet-1K | 86.2M | 418.7G | 80K | 50.2 | 63.0 | [config](configs/sem_fpn/moganet/fpn_moganet_large_80k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_large_80k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/fpn_moganet_large_80k_ade20k.pth) |

### MogaNet + UperNet

| Method | Backbone | Pretrain | Params | FLOPs | Iters | mIoU | mAcc | Config | Download |
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
| UperNet | MogaNet-XT | ImageNet-1K | 30.4M | 855.7G | 160K | 42.2 | 55.1 | [config](configs/upernet/moganet/upernet_moganet_xtiny_512x512_160k_ade20k.py) | log / model |
| UperNet | MogaNet-T | ImageNet-1K | 33.1M | 862.4G | 160K | 43.7 | 57.1 | [config](configs/upernet/moganet/upernet_moganet_tiny_512x512_160k_ade20k.py) | log / model |
| UperNet | MogaNet-S | ImageNet-1K | 55.3M | 946.4G | 160K | 49.2 | 61.6 | [config](configs/upernet/moganet/upernet_moganet_small_512x512_160k_ade20k.py) | log / model |
| UperNet | MogaNet-B | ImageNet-1K | 73.7M | 1050.4G | 160K | 50.1 | 63.4 | [config](configs/upernet/moganet/upernet_moganet_base_512x512_160k_ade20k.py) | log / model |
| UperNet | MogaNet-L | ImageNet-1K | 113.2M | 1176.1G | 160K | 50.9 | 63.5 | [config](configs/upernet/moganet/upernet_moganet_large_512x512_160k_ade20k.py) | log / model |
| UperNet | MogaNet-XT | ImageNet-1K | 30.4M | 855.7G | 160K | 42.2 | 55.1 | [config](configs/upernet/moganet/upernet_moganet_xtiny_512x512_160k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_xtiny_512x512_160k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_xtiny_512x512_160k_ade20k.pth) |
| UperNet | MogaNet-T | ImageNet-1K | 33.1M | 862.4G | 160K | 43.7 | 57.1 | [config](configs/upernet/moganet/upernet_moganet_tiny_512x512_160k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_tiny_512x512_160k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_tiny_512x512_160k_ade20k.pth) |
| UperNet | MogaNet-S | ImageNet-1K | 55.3M | 946.4G | 160K | 49.2 | 61.6 | [config](configs/upernet/moganet/upernet_moganet_small_512x512_160k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_small_512x512_160k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_small_512x512_160k_ade20k.pth) |
| UperNet | MogaNet-B | ImageNet-1K | 73.7M | 1050.4G | 160K | 50.1 | 63.4 | [config](configs/upernet/moganet/upernet_moganet_base_512x512_160k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_base_512x512_160k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_base_512x512_160k_ade20k.pth) |
| UperNet | MogaNet-L | ImageNet-1K | 113.2M | 1176.1G | 160K | 50.9 | 63.5 | [config](configs/upernet/moganet/upernet_moganet_large_512x512_160k_ade20k.py) | [log](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_large_512x512_160k_ade20k.log.json) / [model](https://github.com/Westlake-AI/MogaNet/releases/download/moganet-seg-weights/upernet_moganet_large_512x512_160k_ade20k.pth) |

## Training

Expand Down

0 comments on commit 5a72c32

Please sign in to comment.