Implementation of papers:
- Spatial-Temporal Knowledge-Embedded Transformer for Video Scene Graph Generation
IEEE Transactions on Image Processing (IEEE TIP)，2024.
Tao Pu, Tianshui Chen, Hefeng Wu, Yongyi Lu, Liang Lin
Firstly, we download the directory of data and fasterRCNN in Yrcong' repository.
Then, we follow the instructions to compile some code for bbox operations.
cd lib/draw_rectangles
python setup.py build_ext --inplace
cd ..
cd fpn/box_intersections_cpu
python setup.py build_ext --inplace
For the object detector part, please follow the compilation from https://github.com/jwyang/faster-rcnn.pytorch
@article{Pu2024STKET,
author={Pu, Tao and Chen, Tianshui and Wu, Hefeng and Lu, Yongyi and Lin, Liang},
title={Spatial-Temporal Knowledge-Embedded Transformer for Video Scene Graph Generation},
journal={IEEE Transactions on Image Processing},
volume={33},
pages={556-568},
year={2024},
publisher={IEEE},
doi={10.1109/TIP.2023.3345652}
}
@inproceedings{Pu2023VidSGG,
author={Pu, Tao},
title={Video Scene Graph Generation with Spatial-Temporal Knowledge},
booktitle={Proceedings of the 31st ACM International Conference on Multimedia},
year={2023},
pages={9340--9344},
publisher={Association for Computing Machinery},
doi={10.1145/3581783.3613433}
}
For any questions, feel free to open an issue or contact us: