Raspberry Pi / Computer Vision Engineer (RTSP Human Detection + WhatsApp Alerts + Daily Reporting)
About ForecourtIQ: ForecourtIQ is an AI-powered forecourt analytics platform for car dealerships. We run video analytics on an edge device (Raspberry Pi) to detect on-site activity and generate actionable insights. We’re hiring a developer to build the core “person detection + notification + daily reporting” pipeline. Project Goal: Build production-ready Raspberry Pi software that runs 24/7 and: 1. Watches RTSP camera streams (Google Nest system) 2. Detects when a human enters the forecourt 3. Sends a single WhatsApp notification per visit (no spamming) 4. Counts/tallies human detections/visits 5. Generates a daily end-of-day report This needs to be reliable and resilient in real-world conditions (stream drops, lighting changes, busy scenes, etc.). Key Requirements: Video + Detection: • Ingest live RTSP streams continuously (1+ cameras, scalable) • Run human/person detection on-device • Optimised for Raspberry Pi performance (efficient pipeline) WhatsApp Notification (No Spam): • Send a WhatsApp alert when a human is detected entering the forecourt • Must not repeatedly alert while the person is walking around (e.g., between cars) • Implement anti-spam logic such as AI. • Only alert again when the forecourt has been “clear” for a defined period, or a new visit is detected Daily Reporting: • Maintain counts (e.g., total visits / total detected persons — to be defined in approach) • Produce an end-of-day report automatically (time configurable) • Report output format: Basic text message via WhatsApp. Reliability: • Runs 24/7 as a Linux service (systemd) • Auto-reconnect on RTSP dropouts • Logging + basic health monitoring • Easy configuration and deployment instructions Preferred Tech Stack: • Python (preferred) + OpenCV / GStreamer / FFmpeg • Efficient CV model options: YOLO / TensorFlow Lite / OpenVINO (if applicable to Pi) • Local storage for logs and daily summaries (SQLite is fine) Deliverables: • Working Raspberry Pi application (production-ready) • Documentation (install, config, run, troubleshoot) • Config file for streams + alert rules • Daily report generation and saved outputs • WhatsApp alert integration working end-to-end Nice-to-Have: • Experience with edge AI optimisation on Raspberry Pi • Multi-camera scaling • Zone-based detection (forecourt region-of-interest masking) • Experience with WhatsApp APIs (Twilio WhatsApp, Meta WhatsApp Cloud API, etc.) To Apply (Required) Please include: • Your experience with RTSP + Raspberry Pi + 24/7 services • Which detection approach you’d use (and why) on a Raspberry Pi • How you would implement “notify once per visit” anti-spam logic • Examples of similar work (links or brief descriptions) Generic proposals will be ignored. Apply tot his job