YOLO detects people in ROI · events to HA · Mesh / Wi-Fi IoT linkage
Camera RTSP feeds enter the edge; focus on people and targets inside configured ROI / Zone.
NVIDIA DeepStream real-time pipeline runs YOLO inference — occupied / empty / count and confidence.
Detection events go to Home Assistant to drive Mesh lighting and Wi-Fi devices.
IP camera stream into the edge
NVIDIA GPU-accelerated video pipeline
ROI people / object judgment
Local automation hub receives events
Mesh / Wi-Fi device actions
MXCHIP vertical stack: in-house vision + edge hub + own IoT — project-ready customization and deep secondary development.
DeepStream + YOLO landed and tuned by MXCHIP edge team; ROI / models / scene rules can be customized per project.
Events enter Home Assistant locally; sensor mapping, automations and scenes are delivered by MXCHIP with on-site secondary development.
HA connects MXCHIP Mesh lighting and heterogeneous Wi-Fi products end-to-end — lighting / alert / energy scenes.

Detect people in a clean office and count occupancy in real time.
Outputs occupied / empty / count and confidence; aligned with Ultralytics Region / TrackZone. Events enter HA as local sensors for lighting, curtains, HVAC and more.

Detect vehicles in villa parking / driveway spots
DeepStream multi-stream RTSP inference; YOLO separates enter / leave / stay. Results drive locks, garage lights and barriers via HA (per project).

Detect smoking and similar violations in no-smoking corridors
Person detection plus pose / smoke cues; conclusions stay on-edge and enter HA automations via AI results — no always-on public cloud.

Turn vision events into actionable local alerts
Map detection / behavior to HA binary sensors or the event bus — phone notify, sound/light, locks and Mesh actions: see → judge → act.
Edge DeepStream + YOLO detects people in ROI; HA links IoT — book an advisor for the deployment path.
Book an advisor