CV based Real-Time Bridge People Detection and Visualization

This repository contains Python scripts for real-time people detection and visualization for Bridge Traffic monitoring using the OpenPifPaf library. The project allows real-time pose detection on video streams and provides options to visualize these detections both in the video feed and in a plan view.

Features

• Real-Time Pose Detection: Detects people in real-time video streams from an RTSP source.
• Facial Privacy Option: Provides an option to blur facial features for privacy concerns.
• Plan View Visualization: Visualizes the detected poses on a plan view image, offering a top-down perspective of the scene.

Setup and Installation

Prerequisites

• Python 3.6 or higher
• Ensure you have a working webcam or RTSP video source configured.

Installation

pip install -r requirements.txt

Usage

Real-Time Pose Detection

To run real-time pose detection in the video feed:

python main.py

By default, facial privacy is enabled. You can disable it by modifying the FACIAL_PRIVACY variable in main.py.

Configuration

• main.py: Entry point of the application. Configure RTSP settings, image paths, and other parameters in this file.
• real_time_stream.py: Contains the RealTimeStream class, which handles real-time pose detection and plan view visualization.
• uiuc_my_540_30.png: UIUC banner image for overlay in the video feed.
• bridge_plan_new.png: Plan view image for visualizing detections.
• video_run_v1.avi: Video file for detection visualization.
• sstl_static.png: Static image for visualization in the plan view.
• requirements.txt: Contains the list of required packages and their versions.

Customization

• Modify the RTSP settings in real_time_stream.py to use your specific video source.
• Adjust the plan view coordinates and images according to your setup in real_time_stream.py.

License

This project is licensed under the MIT License.