Counting_With_Adaptive_Auxiliary_Learning

Data Prepare

Prepare you own data, please refer to C-3-FrameWork for detailed instructions.

SHA

• Download checkpoint_best.pth and put it into ./checkpoints/GCN_paper_SHAB/

prediction

python test_SHA.py  

JHU

• Download checkpoint_best.pth and put it into ./checkpoints/GCN_paper_JHU/

prediction

python test_JHU.py

QNRF

• Download checkpoint_best.pth and put it into ./checkpoints/GCN_paper_QNRF/

prediction

python test_QNRF.py

NWPU-Crowd

Our method achieved 76.4 MAE and 327.4 MSE on NWPU-Crowd counting benchmark

Citation

If you find our work useful or our work gives you any insights, please cite:

@article{Meng_2022_Adaptive_Counting,
    author    = {Meng, Yanda and Bridge, Joshua and Wei, Meng and Zhao, Yitian and Qiao, Yihong and Yang, Xiaoyun and Huang, Xiaowei and Zheng, Yalin},
    title     = {Counting with Adaptive Auxiliary Learning},
    journal = {arXiv preprint arXiv:2203.04061},
    year      = {2022},
}