Person Detection for Social Distancing and Safety Violation Alert based on Segmented ROI

Person detection algorithm used is MobileNet SSD with Caffe implementation and the model pre-trained on MS-COCO. Both programs uses OpenCV for image processing and utilizing the DNN module (tested on CPU). The programs later tested on several datasets to prove the concepts.

1. Prerequisites and configurations

All the requirements can be installed via the command:

$ pip install -r requirements.txt

The default input is video located in videos file. To change the program to use camera stream as input, you need to change the configuration from CAMERA = False to CAMERA = True.

Note: All configurations can be changed in the config.py file.

2. Run project

For social distancing program, run:

$ python safety_violation_alert.py

For safety violation alert based on segmented ROI program, run:

$ python safety_violation_alert.py

3. Program output

Social distance monitoring	Safety violation alert based on segmented ROI

4. Accuracy for social distance monitoring

Dataset	TP	TN	FP	FN	%
Oxford Town Centre	11	19	14	4	62.5
PETS2009	14	38	19	5	68
VIRAT	9	4	0	10	56.5

5. Accuracy for safety violation alert based on segmented ROI

Dataset	TP	TN	FP	FN	%
CamNeT	55	58	0	5	95.8

6. References

Mobilenet SSD Caffe

Dataset
MegaPixels: Origins, Ethics, and Privacy Implications of Publicly Available Face Recognition Image Datasets

A Camera Network Tracking (CamNeT) Dataset and Performance Baseline

Publication
Person Detection for Social Distancing and Safety Violation Alert based on Segmented ROI

LICENSE

This project is licensed under the terms of the MIT license.