metadata

license: apache-2.0
title: QuickTrack
sdk: streamlit
emoji: 🏃
colorFrom: gray
colorTo: yellow
pinned: false

Real-Time Object Detection with YOLOv8

A Streamlit-based web application for real-time object detection in videos using YOLOv8. This application supports multiple YOLO models, real-time detection, object tracking, and video processing with annotated output.

Demo

Features

Multiple YOLOv8 model support (Nano to XLarge)
Real-time object detection and tracking
Support for video files and live streams
Unique ID tracking for detected objects
Customizable detection confidence
Color-coded object categories
Downloadable processed videos
Interactive web interface

Installation

Prerequisites

Python 3.8 or higher
CUDA-compatible GPU (optional, but recommended for better performance)

Step 1: Clone the Repository

git clone
cd

Step 2: Create a Virtual Environment (Recommended)

Windows

python -m venv venv
venv\Scripts\activate

Linux/Mac

python3 -m venv venv
source venv/bin/activate

Step 3: Install Dependencies

pip install -r requirements.txt

Usage

Starting the Application

streamlit run app.py

Step-by-Step Guide

Select a Model:
- Choose from available YOLOv8 models in the sidebar
- Models range from Nano (fastest) to XLarge (most accurate)
- Review model details in the expandable section
- Click "Load Selected Model" to download and initialize
Configure Settings:
- Adjust detection confidence using the slider
- Lower values detect more objects but may increase false positives
- Higher values are more selective but might miss some objects
Input Selection:
- Choose between "Video File" or "Live Stream URL"
- For video files: Upload MP4 or AVI format
- For streams: Enter a valid stream URL
Start Detection:
- Click "Start Detection" in the sidebar
- Watch real-time detection with bounding boxes
- Each object gets a unique tracking ID
Download Results:
- Stop detection when finished
- Download button appears automatically
- Processed video includes all annotations

About YOLO Models

Available Models

YOLOv8n (Nano):
- Size: 6.7 MB
- Best for: Real-time applications on CPU
- Speed: ⚡⚡⚡⚡⚡
- Accuracy: ⭐⭐
YOLOv8s (Small):
- Size: 22.4 MB
- Best for: Balanced performance
- Speed: ⚡⚡⚡⚡
- Accuracy: ⭐⭐⭐
YOLOv8m (Medium):
- Size: 52.2 MB
- Best for: Standard detection tasks
- Speed: ⚡⚡⚡
- Accuracy: ⭐⭐⭐⭐
YOLOv8l (Large):
- Size: 87.7 MB
- Best for: High accuracy needs
- Speed: ⚡⚡
- Accuracy: ⭐⭐⭐⭐⭐
YOLOv8x (XLarge):
- Size: 131.7 MB
- Best for: Maximum accuracy
- Speed: ⚡
- Accuracy: ⭐⭐⭐⭐⭐⭐

Model Selection Guide

CPU Only: Use Nano or Small models
GPU Available: Medium to XLarge models recommended
Real-time Needs: Nano or Small models
Accuracy Priority: Large or XLarge models
Balanced: Medium model

Technical Details

Built with Streamlit and OpenCV
Uses Ultralytics YOLOv8 implementation
Supports multiple video codecs
Real-time frame processing and buffering
Unique object tracking with IoU
Color-coded object categories
Frame buffer for smooth video writing

Troubleshooting

Video Not Loading:
- Check file format (MP4/AVI supported)
- Ensure file isn't corrupted
- Try a different video codec
Slow Performance:
- Use a smaller YOLO model
- Reduce input video resolution
- Check GPU availability
Detection Issues:
- Adjust confidence threshold
- Try a larger YOLO model
- Ensure good lighting in video
Download Issues:
- Wait for processing to complete
- Check available disk space
- Try a different browser

Requirements

streamlit>=1.24.0
opencv-python-headless>=4.7.0
torch>=2.0.0
torchvision>=0.15.0
numpy>=1.24.0
ultralytics>=8.0.0
python-dateutil>=2.8.2

Acknowledgments

YOLOv8 by Ultralytics - https://docs.ultralytics.com/models/yolov8/
Streamlit Framework - https://streamlit.io
OpenCV Project - https://docs.opencv.org/4.x/index.html