OUR WORK
Case Studies &
Client Deliveries
Discover the high-end applications, real-time computer vision systems, and custom AI integrations we have developed and successfully delivered.
Proven Client Deliveries
Explore a selection of computer vision applications, machine learning systems, and full-stack solutions built for global clients.
Nighttime Headlight and Road Glare Detection using Computer Vision
Built a computer vision system that detects and segments vehicle headlights and road surface reflections in nighttime driving conditions. Unlike standard approaches that only detect headlights, the system also handles glare reflections on wet road surfaces, making it more robust for real-world night driving scenarios.
Engage-AI — Multimodal Student Engagement Detection System
Built an AI-powered classroom engagement monitoring system using computer vision. The system detects students in lecture videos using YOLO, then runs parallel emotion and action detection engines to compute real-time engagement metrics. Results are displayed on a role-based dashboard for teachers and HODs, with export functionality. Built with Python, FastAPI, and React.
ParkAI — Smart Parking Occupancy Monitor & Analytics Pipeline
Built an AI-powered smart parking occupancy monitoring system using computer vision. The system detects vehicles in CCTV or drone footage using YOLOv8, then runs a custom polygon mask overlap checking algorithm to dynamically identify occupied versus vacant parking slots. Real-time telemetry metrics are computed frame-by-frame and overlaid on the output video feed, tracking total vehicles and occupancy rates. Built with Python, OpenCV, YOLOv8, and PyTorch
PupLink — Real-Time Multi-Dog Detection & Behavior Analysis System
Real-time CV system detecting multiple dogs simultaneously, classifying individual behaviors, and ranking each by activity level in live video. Built with YOLOv8 trained on a custom-annotated dataset. Optimized inference pipeline for low-latency video processing on Raspberry Pi edge hardware. Live streaming via WebRTC with FastAPI backend serving real-time detection results. Pipeline: data annotation → YOLOv8 fine-tuning → evaluation → edge deployment.