nissy-shota / UString

[ACM MM 2020] Uncertainty-based Traffic Accident Anticipation

UString

original repository

environment

CUDA and cuDNN version

cat /usr/local/cuda/version.txt
dpkg -l | grep "cudnn"

feature extraction environment

docker compose build
docker compose up -d
docker compose exec core bash
python3 -m pip install torch==1.4.0+cu100 torchvision==0.5.0+cu100 -f https://download.pytorch.org/whl/torch_stable.html
python3 -m pip install mmcv==0.4.2

mkdir lib
cd lib
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
git checkout v1.1.0  # important!
# Move the downloads from https://drive.google.com/drive/folders/1fbjKrzgXv_FobuIAS37k9beCkxYzVavi into mmdetection
cp -r Cascade\ R-CNN/* ./
# compile & install
python3 -m pip install -v -e .
cd ..
python3 -m pip install mmdetection/
python3 -m pip install Cython==0.29.13
python3 -m pip install pycocotools==2.0.0

inference environment

docker compose build
docker compose up -d
docker compose exec inference bash

python3 -m pip install torch==1.2.0
python3 -m pip install torchvision==0.4.0
python3 -m pip install Cython==0.29.13
python3 -m pip install pycocotools==2.0.0
python3 -m pip install -r requirements.txt

Demo

feature extraction

python3 demo.py --task extract_feature --video_file demo/000821_vis.gif

inference

python3 demo.py --task inference --feature_file demo/000821_vis_feature.npz

visualization

python3 demo.py --task visualize  --video_file demo/000821_vis.gif --result_file demo/000821_result.npz

Note visualization and inference container are same.

reference

UString