Real-Time Vehicle & Object Detection System

A production-quality Computer Vision detection program built with Python, OpenCV, and YOLOv8, capable of detecting 80+ object classes in real time from a webcam, video file, or image.

This project was built to demonstrate expertise in computer vision detection pipelines — a core skill for AI engineering roles.

---

🚀 Features

• ⚡ Real-time detection from webcam or video file at smooth FPS
• 🎯 YOLOv8 nano model — fast inference, small footprint (~6 MB)
• 📦 Detects 80 COCO classes (vehicles, people, animals, objects)
• 🎨 Per-class colour-coded bounding boxes with confidence scores
• 📊 Live FPS counter and detection count HUD overlay
• 🖼️ Image mode — saves annotated output to disk
• 🔧 Uses C++-backed OpenCV routines for high-performance frame processing

---

🛠️ Tech Stack

Layer	Technology
Language	Python 3.10+
Computer Vision	OpenCV (cv2)
Detection Model	YOLOv8 (Ultralytics)
Numerical Processing	NumPy
Backend (OpenCV core)	C++

---

▶️ Setup & Usage

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run on live webcam (default)
python detect.py

# 3. Run on a video file
python detect.py --source path/to/video.mp4

# 4. Run on a static image
python detect.py --source path/to/image.jpg

Note: The YOLOv8 nano model (yolov8n.pt, ~6 MB) downloads automatically on first run.

---

📁 Project Structure

object-detection/
├── detect.py           # Main detection script
├── requirements.txt    # Dependencies
└── README.md           # This file

---

🧠 How It Works

1. Model Load: YOLOv8 is loaded via the Ultralytics API (a high-level wrapper over the C++-optimized PyTorch backend).
2. Frame Capture: OpenCV's VideoCapture (a C++-backed binding) reads frames from the webcam or video file.
3. Inference: Each frame is passed to the YOLO model, which returns bounding boxes, class IDs, and confidence scores.
4. Annotation: Custom draw_detection() uses cv2.rectangle() and cv2.putText() to render results.
5. HUD: FPS is calculated using time.perf_counter() and overlaid with a semi-transparent background for readability.