Computer Vision Projects

Made with ❤️ by Nawaf Rayhan

⭐ If this repo helped you, please give it a star!

A collection of 19 standalone, runnable computer vision projects — YOLO detection & tracking, pose estimation, segmentation, MediaPipe, and classical OpenCV — covering surveillance, retail analytics, safety, productivity, and document processing. Every project lives in its own folder with its own script(s), README.md, and requirements.txt, so you can clone the whole repo or just grab the one folder you need.

---

📌 Projects

Surveillance & Safety

Project	What it does
restricted-zone-intrusion-alert	Alerts when a person's feet enter a defined restricted zone
suspicious-behavior-detection	Pose-based heuristic: flags crouching + fast hand movement
fire-smoke-detection	Real-time fire detection with alarm, + training notebook
fall-detection	Pose-based heuristic: flags a sudden drop into a horizontal posture
ppe-safety-compliance	Flags missing hard hats / safety vests / masks on site
drowsiness-yawn-detection	Driver-monitoring style eye-closure + yawn detection
face-blur-anonymizer	Blurs/pixelates every face in a feed for privacy compliance
rat-rodent-detection	Rodent detection + sighting log, + training notebook

Retail & Crowd Analytics

Project	What it does
people-counter-in-out	IN/OUT footfall counter with a live dashboard
crowd-heatmap	Live crowd density heatmap + person count
queue-wait-time-estimator	Live queue length + rolling average wait time
parking-space-occupancy	Per-slot free/occupied detection for a parking lot

Traffic & Vehicles

Project	What it does
vehicle-detection-tracking	Multi-class vehicle detection with persistent tracking IDs
license-plate-recognition	ANPR: plate detection + OCR reading, + training notebook

Productivity & HCI

Project	What it does
focus-guard-distraction-detector	Self-monitoring screen-distraction guard with full-screen alerts
hand-type-detector	Labels each detected hand Left / Right
hand-gesture-volume-control	Pinch-gesture system volume control

Document & Vision Utilities

Project	What it does
document-scanner-ocr	Classic CV document edge-detect + perspective warp + OCR
instance-segmentation-bg-removal	Green-screen-style background blur/replace via segmentation

---

🛠️ Tech Stack

• Detection & tracking: Ultralytics YOLO (v8/v11) + ByteTrack via supervision
• Pose & hands: YOLO-Pose, MediaPipe, cvzone
• Segmentation: YOLO-Seg
• OCR: EasyOCR, Tesseract
• Core: Python, OpenCV, NumPy
• Training: Google Colab notebooks + Roboflow datasets, for the projects that need a custom-trained model

⚙️ Installation

Each project is self-contained — clone the repo, then install only what that project needs:

git clone https://github.com/Nawaf-Rayhan585/YOLO_Projects.git
cd YOLO_Projects/<project-folder>
pip install -r requirements.txt

▶️ Usage

Every script takes a --source argument (a video file path, an image path, or a webcam index like 0) plus project-specific flags — see each folder's own README.md for the exact command and options. General pattern:

python <script>.py --source your_video.mp4
python <script>.py --source 0              # webcam

📁 Project Structure

Two kinds of projects in this repo:

• Ready to run — most projects use a general-purpose pretrained YOLO/ MediaPipe model and work immediately after pip install.
• Train then run — a few (fire, rodent, license plate, PPE detection) need a model trained on a domain-specific dataset. Those folders include a Colab notebook (train_*.ipynb) that trains one on a free GPU in ~15-30 minutes, plus a local *_inference.py / *_detect.py script to run the result.

No video/image datasets or trained weights are committed to the repo (keeps it light) — point each script's --source/--model flags at your own.

📜 License

MIT — free to use, modify, and build on.

📬 Contact

Built by Nawaf Rayhan — feel free to reach out for freelance CV/AI work.

⭐ Don't forget to star the repo if you found it useful!