Swin Transformer (IGIE)

Model Description

Swin Transformer is a pioneering neural network architecture that introduces a novel approach to handling local and global information in computer vision tasks. Departing from traditional self-attention mechanisms, Swin Transformer adopts a hierarchical design, organizing its attention windows in a shifted manner. This innovation enables more efficient modeling of contextual information across different scales, enhancing the model's capability to capture intricate patterns.

Supported Environments

Model Preparation

Prepare Resources

Pretrained model: https://huggingface.co/docs/transformers/model_doc/swin

git lfs install
git clone https://huggingface.co/microsoft/swin-tiny-patch4-window7-224 swin-tiny-patch4-window7-224

Dataset: https://www.image-net.org/download.php to download the validation dataset.

Install Dependencies

pip3 install -r requirements.txt

Model Conversion

python3 export.py --output swin_transformer.onnx

# Use onnxsim optimize onnx model
onnxsim swin_transformer.onnx swin_transformer_opt.onnx

Model Inference

export DATASETS_DIR=/Path/to/imagenet_val/

FP16

# Accuracy
bash scripts/infer_swin_transformer_fp16_accuracy.sh
# Performance
bash scripts/infer_swin_transformer_fp16_performance.sh

Model Results