Real-Time Chicken Crossing Analytics

I need a compact bot that taps directly into motion-sensor feeds positioned along a rural roadway and turns that stream into usable, real-time analytics about chicken crossing patterns. The goal is not merely to log when a bird is detected; I want the system to recognise multiple chickens, track direction and speed, flag unusual behaviour, and expose the insights through a lightweight dashboard or API I can query from a farm-management app.

You will handle the full pipeline: ingesting raw motion-sensor output, filtering noise, running the analytics logic, and pushing formatted events in under a second. Please keep the solution hardware-agnostic—sensors deliver standard digital pulses at 10 Hz—so swapping in future devices is painless. A small data buffer for brief outages and basic health metrics (uptime, dropped packets) should be included.

Deliverables
• Source code with clear build/run instructions
• Configuration file for sensor endpoints and thresholds
• Live dashboard or documented JSON endpoint showing count, direction, speed, and anomaly alerts
• A read-me that explains how to add new sensors without code changes

Acceptance criteria
1. System processes at least 100 events/second on a Raspberry Pi 4.
2. Dashboard/API updates within one second of any crossing event.
3. Missed-detection rate under 2 % in a 30-minute test with recorded sensor data I will provide.

If you have prior experience with embedded Python, Node-RED, or similar real-time frameworks, mention it. Looking forward to seeing how you can make chicken crossings crystal clear.