Photon — Vision AI for infrastructure.

The Vision AI layer that sees what supervisors cannot.

Photon is the computer vision intelligence layer of the Cognecto platform — applying industrial-grade AI models to existing CCTV infrastructure (RTSP/RTMP/ONVIF) to extract verified safety, quality, and operational events from camera streams in real time.

The capability covers PPE compliance, no-go zone enforcement, fire and smoke detection, driver fatigue (DMS), poor visibility detection, cement spread uniformity, compaction lane coverage, payload estimation, and equipment working/idling classification. Every event is timestamped, geo-tagged, and stored as immutable evidence.

• Existing CCTV — Hikvision, Dahua, Axis, Bosch, ONVIF — all supported
• Industrial AI models trained on dust, rain, snow, low-light worksites
• Real-time inference at the edge — alerts dispatched in 150ms
• Video evidence attached to every event — RCA-ready, court-admissible

Camalco · Hindalco · BEML · Fura Gems · UPRRDA.

Photon is live across mining (Camalco, Hindalco, Fura Gems), manufacturing (BEML), and construction (UPRRDA) — each environment with its own visual complexity, weather conditions, and operational chaos. Industrial AI models are trained for these conditions specifically. PPE compliance reached 97% uniform across three shifts at Hindalco. Near-misses fell 50% at Camalco. UPRRDA Package 6152 achieved 95/100 quality score.

Safety + quality + operations.

Safety: PPE detection, no-go zone intrusion, fire/smoke detection, driver fatigue (DMS), poor visibility alerts. Quality: cement spread uniformity, roller lane coverage, defect classification, edge line continuity. Operations: payload estimation, equipment working/idling, ANPR access control. Each capability is production-ready across customer environments.

The numbers behind the story.