you can look for the 'requirements.txt' to see the requirements
for a new environment you can run below codes
conda create --name boken_test python==3.8.15
conda activate boken_test
conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=10.1 -c pytorch
pip install requirements.txt
dataset download link: https://glass-data.s3.amazonaws.com/Bokeh-Challenge/Ntire23-Bokeh-v7/ntire23-bokeh-v7-train.zip
将前19500图片及其metadata, mask放入 data/train, 将后500张图片及其metadata, mask放入data/val
python train_bokeh.py
Download the pretrained model from Google Drive, and place it in the folder saved_models.
Run the following code to generate test results.
put the video into BKchallenge/huawei_task/huawei_video
#run:
sh test.sh
https://github.com/xuebinqin/U-2-Net
Inference时使用ISNet生成前景分割mask，训练时使用训练集自带的alpha mask
dataset download link: https://glass-data.s3.amazonaws.com/Bokeh-Challenge/Ntire23-Bokeh-v7/ntire23-bokeh-v7-train.zip
将前19500图片及其metadata, mask放入 data/train, 将后500张图片及其metadata, mask放入data/val
python train_bokeh.py
Download the pretrained model from Google Drive, and place it in the folder saved_models.
Run the following code to generate test results.
put the video into BKchallenge/huawei_task/huawei_video
#run:
sh test.sh