This is the repository for S-DCNet, presented in our paper in the ICCV 2019:
From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer
Haipeng Xiong1, Hao Lu2, Chengxin Liu1, Liang Liu1, Zhiguo Cao1, Chunhua Shen2
1Huazhong University of Science and Technology, China
2The University of Adelaide, Australia
- Reformulating the counting problem: We propose S-DCNet, which transforms open-set counting into a closed-set problem via Spatial Divide-and-Conquer;
- Simple and effective: S-DCNet achieves the state-of-the-art performance on three crowd counting datasets (ShanghaiTech, UCF_CC_50 and UCF-QNRF), a vehicle counting dataset (TRANCOS) and a plant counting dataset (MTC). Compared to the previous best methods, S-DCNet brings a 20.2% relative improvement on the ShanghaiTech Part_B, 20.9% on the UCF-QNRF, 22.5% on the TRANCOS and 15.1% on the MTC.
Please install required packages according to requirements.txt
.
Testing data for ShanghaiTech dataset have been preprocessed. You can download the processed dataset from:
Baidu Yun (314M) with code: ou3b
Pretrained weights can be downloaded from:
Baidu Yun (210MB) with code: 1tcb
-
Download the code, data and model.
-
Organize them into one folder. The final path structure looks like this:
-->The whole project
-->Test_Data
-->SH_partA_Density_map
-->SH_partB_Density_map
-->model
-->SHA
-->SHB
-->Network
-->class_func.py
-->merge_func.py
-->SDCNet.py
-->SHAB_main.py
-->main_process.py
-->Val.py
-->load_data_V2.py
-->IOtools.py
-
Run the following code to reproduce our results. The MAE will be SHA: 57.575, SHB: 6.633. Have fun:)
python SHAB_main.py
If you find this work or code useful for your research, please cite:
@inproceedings{xhp2019SDCNet,
title={From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer},
author={Xiong, Haipeng and Lu, Hao and Liu, Chengxin and Liang, Liu and Cao, Zhiguo and Shen, Chunhua},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
year={2019}
}