Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
The key idea of Swin transformer is that the features in shifted window go through transformer module rather than the whole feature map. Besides that, Swin transformer extracts features of different levels. Additionally, compared with Vision Transformer (ViT), the resolution of Swin Transformer in different stages varies so that features with different sizes could be learned. Figure 1 shows the model architecture of Swin transformer. Swin transformer could achieve better model performance with smaller model parameters and less computation cost on ImageNet-1K dataset compared with ViT and ResNet.[1]
Figure 1. Architecture of Swin Transformer [1]
Our reproduced model performance on ImageNet-1K is reported as follows.
| Model | Context | Top-1 (%) | Top-5 (%) | Params (M) | Recipe | Download |
|---|---|---|---|---|---|---|
| swin_tiny | D910x8-G | 80.82 | 94.80 | 33.38 | yaml | weights |
- Context: Training context denoted as {device}x{pieces}-{MS mode}, where mindspore mode can be G - graph mode or F - pynative mode with ms function. For example, D910x8-G is for training on 8 pieces of Ascend 910 NPU using graph mode.
- Top-1 and Top-5: Accuracy reported on the validation set of ImageNet-1K.
Please refer to the installation instruction in MindCV.
Please download the ImageNet-1K dataset for model training and validation.
- Distributed Training
It is easy to reproduce the reported results with the pre-defined training recipe. For distributed training on multiple Ascend 910 devices, please run
# distributed training on multiple GPU/Ascend devices
mpirun -n 8 python train.py --config configs/swintransformer/swin_tiny_ascend.yaml --data_dir /path/to/imagenetIf the script is executed by the root user, the
--allow-run-as-rootparameter must be added tompirun.
Similarly, you can train the model on multiple GPU devices with the above mpirun command.
For detailed illustration of all hyper-parameters, please refer to config.py.
Note: As the global batch size (batch_size x num_devices) is an important hyper-parameter, it is recommended to keep the global batch size unchanged for reproduction or adjust the learning rate linearly to a new global batch size.
- Standalone Training
If you want to train or finetune the model on a smaller dataset without distributed training, please run:
# standalone training on a CPU/GPU/Ascend device
python train.py --config configs/swintransformer/swin_tiny_ascend.yaml --data_dir /path/to/dataset --distribute FalseTo validate the accuracy of the trained model, you can use validate.py and parse the checkpoint path with --ckpt_path.
python validate.py -c configs/swintransformer/swin_tiny_ascend.yaml --data_dir /path/to/imagenet --ckpt_path /path/to/ckpt
Please refer to the deployment tutorial in MindCV.
[1] Liu Z, Lin Y, Cao Y, et al. Swin transformer: Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 10012-10022.