Vision Systems
That See Everything
Real-time detection, depth estimation, multi-camera tracking, and pose analysis — built with YOLO11 and cutting-edge vision transformers.
5 Computer Vision Systems
Depth-Aware Detection System
Real-time object detection + metric depth estimation
Combines YOLO11 object detection with Depth-Anything V2 to produce real-time bounding boxes augmented with accurate metric depth values. Enables precise spatial awareness for robotics and autonomous navigation.
- Metric depth per detection
- Real-time 30fps
- Sub-meter accuracy
Multi-Camera Tracking System
Persistent object tracking across multiple views
ByteTrack-powered system that maintains consistent object IDs as targets move across overlapping and non-overlapping camera fields of view. Built for surveillance and crowd analytics.
- Cross-camera ID persistence
- Multi-view fusion
- Low-latency pipeline
3D Pose Estimation & Action Recognition
Real-time human pose visualization
Uses YOLO11 Pose to extract 17-keypoint skeleton data and classify actions: Walking, Sitting, Falling, Running. Outputs 3D joint coordinates for biomechanical analysis.
- 17-keypoint skeleton
- Action classification
- 3D joint coordinates
Fall Detection System
Real-time safety monitoring for elderly care
Monitors skeletal keypoint trajectories to detect fall events within 200ms. Triggers real-time alerts for caregivers. Designed for deployment in nursing homes and assisted living facilities.
- 200ms detection latency
- Caregiver alerts
- False positive filtering
Thermal Night Vision Detection System
Low-light person detection using thermal imaging
Processes thermal infrared camera feeds to detect and track people in complete darkness. Adapts YOLO11 for thermal domain with custom preprocessing pipeline. Used in perimeter security and search-and-rescue scenarios.
- Works in total darkness
- Thermal domain adaptation
- Perimeter security ready
- Search & rescue deployment