<< return to Pixycam.com

Bean selection. Sorting machine


My name is Albert, I am an industrial engineer from Barcelona. I am new to the world of electronics and an interesting project has emerged that I think could work with pyxycam.
I have to make a machine capable of selecting beans. A selection between good beans (perfectly white) and bad beans (with spots, not white, broken … etc). The beans are relatively small, between 8mm and 12mm.
The question arises if pixycam would be able to detect and recognize the various faults.

What do you think?

Thank you,



Hello Albert,
I’m not sure. It sounds like a challenging problem for Pixy. In general, pure white objects have no color/hue, which Pixy will have problems with. (sorry)



Thank you for your response :blush:
Really, I only need a camera able to do a photo and then post-proceed it via code in Arduino. I don’t need really a good resolution. I think that only comparing a serial of strategic pixels in the photo can be sufficient…
I don’t know if pixy is a good option for this …


You might consider the RGB mode for Pixy2. You don’t need to grab a whole frame and it gives you good control over selection.

The other option is to teach Pixy2 each color of the beans and Pixy2 will detect each bean and tell you it’s location.




Thank you.
My question is, for example… Let’s supose I can teach Pixy2 how a perfect bean is. If I show it a bean that has a spot, will it be detect as a bean?

There are infinite combinations of a bad bean. I think it’s not possible to teach it all of them.


Pixy can detect and/or distinguish between objects based on hue/color. It doesn’t sound like Pixy can do what you describe. (sorry)