Awesome

Locating_Counting_with_a_Depth_Prior

[TPAMI] Locating and Counting Heads in Crowds With a Depth Prior [Paper]

Dataset

Download or generate the virtual dataset from the ShanghaiTechRGBDSyn repository.

Download ShanghaiTechRGBD dataset from OneDrive.

Depth Completion

Style Transfer

By using SG-GAN.

Train

CUDA_VISIBLE_DEVICES=0,1,2,3 python ./tools/train_GTAV_metric.py \
	--dataset GTAV --dataroot /group/crowd_counting/GTAV-ours/ \
	--cfg_file lib/configs/resnext50_32x4d_GTAV --lr 0.05 --batchsize 8

If you have high capacity GPUs, we recommend training with larger size images.

Inference

python ./tools/test_RGBD_metric.py \
	--dataset RGBD --dataroot /p300/data/Dataset/SIST_RGBD/RGBDmerge_540P/Part_A/ \
	--cfg_file lib/configs/resnext50_32x4d_GTAV \
	--load_ckpt checkpoint.pth

[checkpoint 270x480]

Acknowledgements

This repository borrows partially from VNL and MAML.

Citation

If you find this repository useful for your research, please use the following:

@article{lian2021locating,
  title={Locating and Counting Heads in Crowds With a Depth Prior},
  author={Lian, Dongze and Chen, Xianing and Li, Jing and Luo, Weixin and Gao, Shenghua},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2021},
  publisher={IEEE}
}