LCN (Lightweight Crowd-corunting Net) by using NWPU-Crowd dataset

---

This repo is based on NWPU-Crowd-Sample-Code.

Getting Started

Preparation

• Prerequisites

  • Python 3.x
  • Pytorch 1.x: http://pytorch.org .
  • other libs in requirements.txt, run pip install -r requirements.txt.

• Installation

  • Clone this repo:

    git clone https://github.com/sailyung/crowd_counting.git
    

• Data Preparation

  • Download NWPU-Crowd dataset from this link.
  • Unzip *zip files in turns and place images_part* into a folder. Finally, the folder tree is below:

    -- NWPU-Crowd
        |-- images
        |   |-- 0001.jpg
        |   |-- 0002.jpg
        |   |-- ...
        |   |-- 5109.jpg
        |-- jsons
        |   |-- 0001.json
        |   |-- 0002.json
        |   |-- ...
        |   |-- 5109.json
        |-- mats
        |   |-- 0001.mat
        |   |-- 0002.mat
        |   |-- ...
        |   |-- 5109.mat
        |-- train.txt
        |-- val.txt
        |-- test.txt
        |-- readme.md
  

  • Run ./datasets/prepare_NWPU.m using Matlab.
  • Modify __C_NWPU.DATA_PATH in ./datasets/setting/NWPU.py with the path of your processed data.

the folder tree is below:

Training

• Set the parameters in config.py and ./datasets/setting/NWPU.py (if you want to reproduce our results, you are recommended to use our parameters in ./saved_exp_para).
• run python train.py.
• run tensorboard --logdir=exp --port=6006.

Testing

We only provide an example to forward the model on the test set. You may need to modify it to test your models.

• Modify some key parameters in test.py:
  • Line 32: LOG_PARA, the same as __C_NWPU.LOG_PARA in ./datasets/setting/NWPU.py.
  • Line 34: dataRoot, the same as __C_NWPU.DATA_PATH in ./datasets/setting/NWPU.py.
  • Line 36: model_path.
  • Line 48: GPU Id and Model Name.
• Run python test.py.

Performance on the validation set

The overall results on val set:

Method	O_MAE	O_MSE	O_NAE	Avg.MAE[S]	Avg.MAE[L]
MCNN [1]	232.5	714.6	1.063	1171.9	220.9
SFCN+ [2]	105.7	424.1	0.254	712.7	106.8
LCN	233.097	802.123	0.848	1503.09	218.682

O_MAE: Mean Absolute Error (MAE) on overall testing images. O_MSE: Mean Squared Error (MAE) on overall testing images. O_MAE: Normalized Absolute Error (NAE) on overall testing images. Avg.MAE[S]: The test set is divided into five scene levels according to the numbers of people in an image, and avg. MAE[S] represents the average MAE of the five scene levels. Avg.MAE[L]: The test set is divided into four luminance levels according to the luminance calculated in an image, and avg. MAE[S] represents the average MAE of the four luminance levels.

About the leaderboard on the test set, please visit Crowd benchmark.

References

1. Single-Image Crowd Counting via Multi-Column Convolutional Neural Network, CPVR, 2016.
2. Learning from Synthetic Data for Crowd Counting in the Wild, CVPR, 2019.

Evaluation Scheme

The Evaluation Python Code of the crowdbenchmark.com is shown in ./misc/evaluation_code.py, which is similar to validation code in trainer.py.