Real-Time Object Detection & Tracking System

Computer Vision AI YOLOv8 Deep SORT Streamlit

Overview

This project is an AI-powered object detection and tracking system built using YOLOv8 and Deep SORT. The application can detect and track multiple objects in images, uploaded videos, and real-time webcam streams. A Streamlit-based interface provides an easy-to-use platform for performing object detection, assigning unique IDs to detected objects, and monitoring their movement across frames.

Key Features

Object detection using YOLOv8 deep learning model
Image upload and object recognition
Video-based object detection and analysis
Real-time webcam object detection
Multi-object tracking using Deep SORT
Unique ID assignment for tracked objects
Bounding box visualization and object labeling
Interactive Streamlit web interface
Support for multiple image and video formats
Real-time processing and visualization

Tech Stack

Python

YOLOv8

Ultralytics

Deep SORT

OpenCV

NumPy

Streamlit

Pillow (PIL)

ImageIO

Computer Vision

How It Works

Input Capture: Users upload an image, video, or provide a live webcam feed for analysis.
Object Detection: The YOLOv8 model scans each frame and identifies objects present in the scene.
Object Classification: Detected objects are labeled with their corresponding categories such as person, car, bicycle, and more.
Object Tracking: Deep SORT assigns unique IDs and tracks detected objects across multiple frames.
Visualization: Bounding boxes, labels, and tracking IDs are displayed in real time through the Streamlit interface.

Implementation Details

Model: YOLOv8 pre-trained on COCO dataset
Input: Video files, webcam streams, or image sequences
Processing: Frame-by-frame object detection with tracking
Output: Annotated video with bounding boxes and class labels
Performance: ~30-60 FPS on GPU (RTX 3060+)

Use Cases

Security and surveillance systems
Traffic monitoring and vehicle tracking
Retail analytics and people counting
Industrial quality control
Crowd monitoring and public safety

Project Links

View Code Documentation