https://github.com/gjy3035/pcc-net
PCC Net: Perspective Crowd Counting via Spatial Convolutional Network
https://github.com/gjy3035/pcc-net
computer-vision crowd-analysis crowd-counting multi-task-learning
Last synced: about 1 year ago
JSON representation
PCC Net: Perspective Crowd Counting via Spatial Convolutional Network
- Host: GitHub
- URL: https://github.com/gjy3035/pcc-net
- Owner: gjy3035
- Created: 2018-12-20T02:00:13.000Z (over 7 years ago)
- Default Branch: ori_pt1_py3
- Last Pushed: 2020-03-12T11:12:23.000Z (about 6 years ago)
- Last Synced: 2025-03-24T21:39:12.249Z (about 1 year ago)
- Topics: computer-vision, crowd-analysis, crowd-counting, multi-task-learning
- Language: Python
- Size: 676 KB
- Stars: 76
- Watchers: 6
- Forks: 18
- Open Issues: 18
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# PCC Net: Perspective Crowd Counting via Spatial Convolutional Network
This is an official implementation of the paper "PCC net" (PCC Net: Perspective Crowd Counting via Spatial Convolutional Network).
In the paper, the experiments are conducted on the three populuar datasets: Shanghai Tech, UCF_CC_50 and WorldExpo'10. To be specific, Shanghai Tech Part B contains crowd images with the same resolution. For easier data prepareation, we only release the pre-trained model on ShanghaiTech Part B dataset in this repo.
## Bracnhes
1. [ori_pt0.2_py2](https://github.com/gjy3035/PCC-Net/tree/ori_pt0.2_py2): the original version.
2. [ori_pt1_py3](https://github.com/gjy3035/PCC-Net): the current version.
3. [vgg_pt1_py3](https://github.com/gjy3035/PCC-Net/tree/vgg_pt1_py3): vgg-backbone PCC Net (higher performance).
## Requirements
- Python 3.x
- Pytorch 1.x
- TensorboardX (pip)
- torchvision (pip)
- easydict (pip)
- pandas (pip)
## Data preparation
1. Download the original ShanghaiTech Dataset [Link: [Dropbox ](https://www.dropbox.com/s/fipgjqxl7uj8hd5/ShanghaiTech.zip?dl=0)/ [BaiduNetdisk](https://pan.baidu.com/s/1nuAYslz)]
2. Resize the images and the locations of key points.
3. Generate the density maps by using the [code](https://github.com/aachenhang/crowdcount-mcnn/tree/master/data_preparation).
4. Generate the segmentation maps.
We also provide the processed Part B dataset for training. [[Link](https://mailnwpueducn-my.sharepoint.com/:u:/g/personal/gjy3035_mail_nwpu_edu_cn/EcMLqr9XuH1ChAgkqpxL_6kBK9EyCmIuXMxTb09FrjMYow?e=LJnOcC)]
## Training model
1. Run the train_lr.py: ```python train_lr.py```.
2. See the training outputs: ```Tensorboard --logdir=exp --port=6006```.
In the experiments, training and tesing 800 epoches take 21 hours on GTX 1080Ti.
## Expermental results
### Quantitative results
We show the Tensorboard visualization results as below:
The mae and mse are the results on test set. Others are triaining loss.
### Visualization results
Visualization results on the test set as below:
Column 1: input image; Column 2: density map GT; Column 3: density map prediction; Column 4: segmentation map GT; Column 5: segmentation map prediction.
## Citation
If you use the code, please cite the following paper: