// the find
HectorTa1989/Intruder-Alert-system-with-Azure-IOT-Hub
Intruder alerting system with Azure IOT Hub based on ESP-32 IP Camera
An ESP32-cam sketch that streams video, does basic face detection, and pipes a base64-encoded snapshot to Azure IoT Hub via MQTT when a face is detected. Azure Functions then write it to Blob Storage and send an alert email. Weekend project territory — not a production security system.
- End-to-end pipeline is complete: device → IoT Hub → Function → Blob → email, which is more than most 'hello IoT' demos get to
- Using MQTT directly from the ESP32 is the right transport choice for constrained hardware; avoids HTTP overhead
- Separating blob storage from email delivery into two Functions keeps each piece independently replaceable
- Base64-encoding camera frames on an ESP32 is a bad idea — it inflates payload size by ~33% on a device with 520KB RAM, and MQTT has a 256KB default message size limit on IoT Hub; large images will silently fail or get truncated
- Zero stars, no commits since March 2023, and the README admits the face recognition is copy-pasted from Arduino IDE examples — the 'face recognition' is Espressif's basic haar-cascade demo, not anything you'd trust for real detection
- No error handling documented anywhere: what happens when the MQTT send fails, IoT Hub is unreachable, or the Function throws? No retry logic, no dead-letter queue
- Hardcoded credentials are almost certainly in the .ino file given the project structure — no mention of secrets management, certificates, or SAS token rotation