Tianyu Wang*, Xiaowei Hu*, Qiong Wang, Pheng-Ann Heng, and Chi-Wing Fu (* Joint first authors.)
[openaccess][arXiv] [BibTeX]
Instance shadow detection aims to find shadow instances paired with object instances. We present a dataset, a deep framework, and an evaluation metric to approach this new task. This repo is implemented on Detectron2.
Install LISA and pysobatools
$ cd InstanceShadowDetection $ python setup.py install $ cd PythonAPI $ python setup.py install
$ cd InstanceShadowDetection/docker $ docker build --network=host --tag="instanceshadow" -f ./Dockerfile . $ docker run --gpus all -it --ipc=host --name=instanceshadow --network=host -v /YOURPATH:/data instanceshadow:latest
(Nvidia-docker)[https://github.com/NVIDIA/nvidia-docker] is needed.
Please dowload from Google Drive. Put the model in projects/LISA/output_light/.
$ cd projects/LISA/
$ python demo.py --input ./demo/web-shadow0573.jpg --output ./ --config ./config/LISA_101_FPN_3x_demo.yaml$ python train_net.py --num-gpus 2 --config-file ./config/LISA_101_FPN_3x.yaml
$ python train_net.py --num-gpus 2 --config-file ./config/LISA_101_FPN_3x.yaml --eval-only --resume $ python SOAP.py
python visualize_json_results.py --ins_input ./output_light/inference/soba_instances_results.json --ass_input ./output_light/inference/soba_association_results.json --output ./output_light/results --dataset soba_cast_shadow_val_full
If you use LISA, SOBA, or SOAP, please use the following BibTeX entry.
@InProceedings{Wang_2020_CVPR,
author = {Wang, Tianyu and Hu, Xiaowei and Wang, Qiong and Heng, Pheng-Ann and Fu, Chi-Wing},
title = {Instance Shadow Detection},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2020}
}