Skip to content

Latest commit

 

History

History
70 lines (54 loc) · 3.5 KB

README.md

File metadata and controls

70 lines (54 loc) · 3.5 KB

Vehicle Tracking and ROI Timer Application

This C++ application leverages computer vision and deep learning to track vehicles in a video and implement a timer within a designated Region of Interest (ROI). The output is a processed video with bounding boxes, ROI visualization, and wait times displayed for each vehicle within the ROI.

Features

  • Vehicle Detection: Employs a YOLOv8 model for real-time vehicle identification within the video stream.
  • Vehicle Tracking: Tracks vehicles across frames to maintain identity and calculate wait times.
  • ROI Timer: Implements a timer for each vehicle present within the user-defined ROI.
  • Video Output: Generates a processed video with bounding boxes, ROI, and wait time overlays.

Project Structure

  • CMakeLists.txt: CMake build configuration.
  • Dockerfile: Instructions for building a Docker container for the application.
  • README.md: This documentation file.
  • build.sh: Shell script for building the application.
  • docs: Directory for additional documentation.
  • input: Directory containing input videos and model configuration files.
  • logs: Directory for application logs.
  • models: Directory for storing the YOLOv8 model file.
  • output: Directory for storing the processed output video.
  • run.sh: Shell script for running the application.
  • src: Directory containing the C++ source code.

Program Flow

  1. Initialization:

    • Input Handling: Reads command-line arguments or configuration to get the input video path, output video path, model path, classes file, and ROI coordinates.
    • Video Capture: Opens the input video file using OpenCV's VideoCapture.
    • Video Writer: Creates an output video file using OpenCV's VideoWriter.
    • Model Loading: Loads the YOLOv8 object detection model.
    • Vehicle Tracker Initialization: Sets up the vehicle tracking system.
    • ROI Definition: Processes the ROI coordinates and establishes the ROI region.

  2. Frame Processing Loop:

    • Frame Reading: Reads a frame from the input video.
    • ROI Cropping: Crops the ROI region from the frame.
    • Vehicle Detection: Performs vehicle detection on the ROI-cropped frame using the YOLOv8 model.
    • Vehicle Tracking: Updates the vehicle tracker with the detected vehicles.
    • Timer and Overlay: Calculates wait times for vehicles within the ROI and overlays bounding boxes, ROI, and wait times on the frame.
    • Frame Writing: Writes the processed frame to the output video.

  3. Termination:

    • Cleanup: Releases video capture and writer objects.
    • Logging: Logs any errors or exceptions encountered during processing.

Usage

  1. Build:

    • Execute build.sh to Build the Docker image using the provided Dockerfile and to compile the C++ code and generate the executable.

  2. Run:

    • Execute run.sh providing the required arguments, to run the application within the Docker container, mapping the necessary volumes for input and output files:
      • input_video_path: Path to the input video file.
      • output_video_path: Path to save the processed output video.
      • model_path: Path to the YOLOv8 model file.
      • classes_file: Path to the file containing class names (e.g., coco.names).
      • roi_coordinates: ROI coordinates in the format "x,y,width,height".

Dependencies

  • OpenCV
  • C++17
  • Docker

Note

  • Ensure you have the required dependencies installed.
  • Adjust the paths in the scripts and code according to your project structure.