Skip to content

Latest commit

 

History

History
51 lines (36 loc) · 1.9 KB

README.md

File metadata and controls

51 lines (36 loc) · 1.9 KB
Raw

Pedestrian Detection Project

This project demonstrates real-time pedestrian detection using the YOLOv4-tiny object detection model with OpenCV. It processes video frames, detects pedestrians, and annotates them with bounding boxes.

Prerequisites

Before running the script, ensure you have the following installed:

  • Python 3.x
  • OpenCV (pip install opencv-python)
  • NumPy (pip install numpy)
  • imutils (pip install imutils)

Setup

  1. Clone the repository:

    git clone https://github.com/yourusername/pedestrian-detection.git
cd pedestrian-detection

2.Download YOLOv4-tiny model files:

  • Download yolov4-tiny.weights and yolov4-tiny.cfg from the official YOLO website or other sources.
  • Save these files into the project directory.

3.Obtain the COCO labels file (coco.names):

  • The coco.names file contains a list of class names recognized by the YOLO model, including "person".
  • Place coco.names in the project directory.

Usage

1.Run the pedestrian detection script:

python pedestrian_detection.py
  1. Script Behaviour
  • The script opens a video file and starts processing frames for pedestrian detection.
  • Detected pedestrians are annotated with bounding boxes.
  • Press the Esc key to exit the application.

Customization

  • Input Video: Modify the video_path variable in the script (pedestrian_detection.py) to use a different video file.
  • Detection Parameters: Adjust NMS_THRESHOLD and MIN_CONFIDENCE constants in the script for controlling detection sensitivity.
  • Model Configuration: Customize the YOLO model by changing the configPath and weightsPath variables in the script.

Acknowledgments

  • This project uses YOLOv4-tiny, a deep learning model developed by the Darknet team. Visit the official YOLO website for more information.
  • The COCO dataset and labels (coco.names) are used for object recognition. Refer to the COCO website for dataset details.