YOU CAN REPLICATE THIS PROJECT -> https://github.com/Nagharjun17/Fault-DetectionâWiFi-Extender
- Used a 13 MP AR1335 camera pointed at the Wi-Fi extender LED to detect power loss.
- Trained a MobileNetV2 classifier on LED on vs off images using PyTorch and timm on my laptop.
- Exported the best model to ONNX and ran real time inference on the Kria KV260 using ONNX Runtime and the SmartCam RTSP stream.
- Added simple temporal filtering and thresholds so short flickers do not trigger false alarms.
- Published LED state and class probabilities to MQTT topics (via HiveMQ Cloud) so any service can subscribe and alert.