Human Attribute Recognition System (HARS)

Human Attribute Recognition System (HARS)

Human Attribute Recognition System (HARS) is an AI-powered computer vision platform developed to recognize and classify human attributes from images and live video streams using deep learning models. The project focuses on extracting meaningful visual characteristics to support intelligent surveillance analytics and automated monitoring.

The system accurately identifies multiple human attributes including gender, skin tone, approximate age, and clothing style. It recognizes a wide range of dress codes such as traditional Arab attire, Indian clothing, sportswear, business attire, and branded uniforms including Talabat delivery uniforms.

Key Features

• Human detection and recognition using deep learning.
• Gender classification.
• Skin tone classification.
• Approximate age estimation.
• Dress code and clothing style recognition.
• Recognition of traditional, sports, business, and branded uniforms.
• Real-time image and video stream analysis.
• High-accuracy AI classification models.

My Role

I independently led the entire project from concept to deployment. My responsibilities included dataset collection and preparation, data annotation, model training, performance optimization, validation, and system integration. I developed and refined the AI models to achieve high recognition accuracy across multiple human attributes while ensuring reliable real-time performance.

Technologies

Python, PyTorch, OpenCV, Deep Learning, Computer Vision, Machine Learning, Neural Networks, Image Processing, AI Model Training.