FarhatBuet14 / CrowdNET

CrowdNET

CrowdNET

CrowdNET - A Case Study Towards Designing Context- driven Computer Vision Techniques to Identify Anomalous Crowd Behavior in Hajj Pilgrimage

We have designed two computer- vision algorithms to detect physical barriers in any image/ video frame, and to count the number of people therein. By employing two algorithms, we overcome data sparsity challenges in number of images of people climbing on gates to train

Requirements

• Python 3.6.12

Environment Setup

git clone https://github.com/FarhatBuet14/CrowdNET.git
cd CrowdNET
pip install -r requirements.txt

Barrier Detection and Localization

Test with images

• Download the model file from Google Drive. Put the model file on Barrier_Detection_Localization/output folder.
• Put the test images on the Barrier_Detection_Localization/test_images folder

cd Barrier_Detection_Localization/codes
python3 test_image.py

• You will get the result images on the Barrier_Detection_Localization/output/test_images_results folder.

Crowd Estimation

Test with images

• Download the model file from Google Drive. Put the model file on Crowd_Estimation/model folder.
• Put the test images on the Crowd_Estimation/test_images folder

cd Crowd_Estimation/codes
python3 test.py

• You will get the result images on the Crowd_Estimation/output folder.