detector

Video Object Detection with PaliGemma

This project demonstrates how to perform object detection on video frames using the google/paligemma-3b-mix-224 model. The script processes an input video, detects objects specified in a prompt, and draws bounding boxes around detected objects in the output video.

Table of Contents

• Installation
• Usage
• Project Structure

Installation

1. Clone the repository:

   git clone https://github.com/your-username/video-object-detection.git
   cd video-object-detection

2. Create and activate a virtual environment (optional but recommended):

   python3 -m venv venv
   source venv/bin/activate

3. Install the required packages:

   pip install -r requirements.txt

4. Log in to Hugging Face:

   !huggingface-cli login

Usage

1. Place your input video file in the project directory and name it input_video.mp4. You can also change the input video filename in the script if needed.

2. Run the script:

   python video_object_detection.py

3. The processed video with bounding boxes will be saved as output_video.avi.

Project Structure

video-object-detection

input_video.mp4 # Input video file

output_video.avi # Output video file (generated)

video_object_detection.py# Main script for object detection

requirements.txt # List of dependencies

README.md # Project README file