Vehicle Detection + Tracking App

GitHub

Overview

This repository contains a Streamlit web application for vehicle tracking using different SOTA object detection models. The app offers two options: YOLO-NAS with SORT tracking and YOLOv8 with ByteTrack and Supervision tracking. It enables users to upload a video file, set confidence levels, and visualize the tracking results in real-time.

Technologies Used

• Python
• Streamlit
• OpenCV
• PyTorch
• YOLO-NAS
• YOLOv8
• SORT
• ByteTrack
• Supervision

How to Clone and Run the App

1. Clone the Repository

   git clone https://github.com/your-username/vehicle-tracking-app.git
   

2. Install Dependencies

   pip install -r requirements.txt
   

3. Run the App

   streamlit run app.py
   

Object Detection and Tracking Overview

YOLO-NAS with SORT Tracking

• YOLO (You Only Look Once) is a real-time object detection system. YOLO-NAS (Neural Architecture Search) is a YOLO variant that is optimized for detecting a wide range of objects.
• SORT (Simple Online and Realtime Tracking) is a simple, yet effective, online multiple object tracking algorithm. It associates detections in consecutive frames to track objects.

YOLOv8 with ByteTrack and Supervision

• YOLOv8 is another variant of YOLO, known for its high accuracy and speed in object detection.
• ByteTrack is a state-of-the-art online multi-object tracking algorithm that employs motion and appearance cues to track objects effectively.
• Supervision is a toolkit for YOLO-based object detection models.

YOLO-NAS Overview

• YOLO-NAS, short for You Only Look Once with Neural Architecture Search, is a cutting-edge object detection model optimized for both accuracy and low-latency inference.
• Developed by Deci, YOLO-NAS employs state-of-the-art techniques like Quantization Aware Blocks and selective quantization for superior performance.
• It sets a new standard for state-of-the-art (SOTA) object detection, making it an ideal choice for a wide range of applications including autonomous vehicles, robotics, and video analytics.

Unique Features of YOLO-NAS

• Utilizes Quantization Aware Blocks for efficient inference without sacrificing accuracy.
• Incorporates AutoNAC technology for optimal architecture design, balancing accuracy, speed, and complexity.
• Supports INT8 quantization for unprecedented runtime performance.
• Pre-trained weights available for research use on SuperGradients, Deci’s PyTorch-based computer vision training library.

YOLOv8 Overview

• YOLOv8 is a state-of-the-art object detection model developed by Ultralytics, known for its high accuracy and developer-friendly features.
• It introduces significant architectural improvements over its predecessors and is actively supported by the community.

Main Features of YOLOv8

• Utilizes an anchor-free detection system for more accurate predictions.
• Incorporates new convolutional blocks for improved performance.
• Implements Mosaic Augmentation for enhanced training.

ByteTrack and SORT Tracking

• ByteTrack and SORT are advanced online multi-object tracking algorithms that complement YOLO-based object detection models.
• ByteTrack leverages motion and appearance cues for effective object tracking.
• SORT associates detections in consecutive frames to create object tracks.