TIAM62AITrafficLight

AI traffic lights with Texas Instruments AM62A and Edge Impulse.

A model is trained to recognize motorcycle riders with and without helmets. That model is loaded into TI AM62A. A regular webcam will send pictures for local inference. Inference values will be sent to a server.

A separate WiFi device - Unihiker with Debian/Python - will query the server to determine if red traffic light will switch to green.

Parts

Texas Instruments AM62A Unihiker board USB Webcam Web server/PHP

Tutorial

https://docs.edgeimpulse.com/experts/image-projects/motorcycle-helmet-detection-smart-light-ti-am62a

Machine Learning model

https://studio.edgeimpulse.com/public/295740/latest (Trained with Lego)

3d printed part

https://www.thingiverse.com/thing:6277702

Spanish info

https://bandini.medium.com/sem%C3%A1foro-con-detecci%C3%B3n-de-cascos-v%C3%ADa-machine-learning-15a18f60d218

Demo

https://www.youtube.com/watch?v=YR2Mjf_unmU

Contact

@RoniBandini