Triad Spectroscopy AS7265x Sensors, fittable for material identification?

Hey, I am an AI student, and I am building this project of a smart trash bin that detects the type of trash and processes it. I have stumbled upon the triad spectroscopy sensor and am thinking if it could be of use for my project.

I need to fuel the labeled test data for my model with Plastic, Metal, Drink cartons, Glass, Paper, Organics. I realize that in theory this would be applicable but I’m a bit worried as some materials would have too close values for the model to attribute it to a certain class.

Btw, I am planning to fuse it with Computer Vision to easen the process up.

Will these sensors do their job in the material identification niche I’m looking for?

The spectroscopy sensor separates the light reflected from a piece of trash and quantifies the intensity in 18 wavelength bins.

Those data could assist in trash identification, so give ML techniques a try, and let us know how well it works.

Thanks for the answer:) I will try to make it work. I hope it all goes smooth. I currently already own a raspi 5 so connecting via I2C shouldn’t be a problem, but I am a bit worried about the libraries. If I find out how to adapt them to the pi infrastructure I’ll definitely give the sensor a shot!

Differentiating classes of items in an environment as diverse as single stream household trash is a big problem to solve. Light spectrum is one characteristic that can be used for sorting. There are many other characteristics that need to be similarly considered but visible color is as good a place to start as any.

This https://github.com/Philliec459/Science- … ectrometer is the best demo of what you can do with it; so I’d say it is possible with enough effort :wink: