A camera that only records is an insurance policy. A camera that detects is an operations tool. CatchupGroup builds computer vision systems that understand what they see — on your existing cameras and your own hardware.
Capabilities
- Detection: people, vehicles, animals and custom objects with YOLO-family models
- Events: line crossing, zone intrusion, loitering, direction and speed
- ANPR: number plate recognition for gates, yards and logistics
- Tracking and counting: throughputs, dwell times and occupancy metrics
- Custom models: trained on your site, your angles, your conditions
How we run it
- RTSP ingest from existing CCTV — most installations keep their cameras
- NVIDIA DeepStream and TensorRT pipelines on edge devices or a private GPU server
- Alerts that mean something: verified detections to Telegram, WhatsApp, MQTT or your control room
- Tuning against false alarms — the difference between a system staff trust and one they mute
See it on your own footage with an Edge AI Vision Kit or a pilot on one camera line first.