Sight: Computer Vision Object Detection 👁️

A practical computer vision project implementing real-time object detection using YOLOv5. This project demonstrates hands-on experience with modern CV pipelines and serves as a foundation for more specialized vision applications.

🚀 Features

• Real-time Object Detection: Identifies common objects (people, animals, vehicles, etc.) using YOLOv5
• Image Processing Pipeline: Handles image input, model inference, and result visualization
• Confidence Scoring: Provides detection confidence levels for each identified object
• Modular Architecture: Clean, extensible codebase for future computer vision projects

🛠️ Tech Stack

• Object Detection: YOLOv5
• Computer Vision: OpenCV
• Deep Learning Framework: PyTorch
• Core Language: Python 3.9+

📸 Demonstration

The YOLOv5 model successfully detects and labels objects in images with confidence scores:

YOLOv5 detecting a person (56% confidence) and tie (40% confidence) in test image.

⚡ Quick Start

Installation

1. Clone the repository:

   git clone https://github.com/Auction-s/Sight.git
   cd Sight
   python -m venv cv_env

Windows

cv_env\Scripts\activate

Linux/Mac

source cv_env/bin/activate pip install -r requirements.txt

🔬 Technical Implementation Core Components YOLOv5 Integration: Pre-trained COCO dataset model for general object detection

OpenCV Pipeline: Image loading, processing, and annotation

Confidence Thresholding: Configurable detection sensitivity

Result Visualization: Clean bounding boxes and labels on output images

Detection Capabilities The current model can identify 80+ common object categories including:

People, animals, vehicles

Everyday items like ties, chairs, cups

Electronic devices, furniture, food items

📚 Lessons Learned Building this computer vision system provided valuable experience with:

Model Deployment: Successfully implemented and tested YOLOv5 for practical object detection

OpenCV Integration: Gained experience with image processing and annotation pipelines

Confidence Interpretation: Learned to evaluate and interpret model confidence scores

Project Setup: Established proper Python project structure with dependency management

Computer Vision Fundamentals: Understood the end-to-end process of object detection systems

🚀 Future Enhancements Custom Training: Fine-tune YOLOv5 on domain-specific datasets

Real-time Video: Extend to video streams and webcam input

Web Interface: Create a Streamlit dashboard for easy testing

Performance Optimization: Implement model quantization for faster inference

Multi-model Comparison: Test different object detection architectures

Specialized Applications: Adapt for specific use cases (retail, security, etc.)