데모 프로그램 소개
가상환경 생성
python3.8 -m venv venv
source ./venv/bin/activate
pip install -U pip
- CUDA11.8
- Python3.8
- pytorch 2.0.0+cu118
- torchvision 0.15.1+cu118
mmaction 관련 패키지 설치
pip install -U openmim
mim install mmengine
mim install "mmcv==2.1.0"
mim install mmdet
Project build
git clone "this repository" project
cd project
pip install -v -e .
Video input Video output download link (weight file) : i3d_imagenet-pretrained-r50-heavy_8xb8-32x2x1-100e_kinetics400-rgb_trained_by_hnu_epoch_40.pth download link (demo video, dump) : i3d_imagenet-pretrained-r50-heavy_8xb8-32x2x1-100e_kinetics400-rgb_trained_by_hnu_epoch_40.pth
python demo/demo_visualize.py configs/recognition/i3d/i3d_dense_trained_by_hnu.py \
i3d_imagenet-pretrained-r50-heavy_8xb8-32x2x1-100e_kinetics400-rgb_trained_by_hnu_epoch_40.pth \
demo/167-3_cam02_dump02_place04_day_summer.mp4 tools/data/kinetics/label_map_hnu_label.txt \
--out-filename output.mp4
 |
 |
|---|---|
| 투기행위 신뢰도 : 0.78 (야간) | 투기행위 신뢰도 : 0.98 (주간) |
 |
 |
| 투기행위 신뢰도 : 0.91 (주간) | 투기행위 신뢰도 : 0.92 (주간) |
- MMDetection: OpenMMLab detection toolbox and benchmark.
- MMAction2: OpenMMLab Video Action Recognition toolbox and benchmark.