I'm interested in exploring machine learning on Arduino, specifically for computer vision applications. However, I'm unsure if Arduino boards have sufficient processing power and memory to handle image data directly.
Can anyone share their experiences or insights on:
Which Arduino boards are best suited for image processing and machine learning tasks?
What are the limitations and challenges when working with image data on Arduino?
Are there any specific libraries, shields, or techniques that can help optimize image processing and machine learning on Arduino?
Are there any successful projects or examples of Arduino-based machine learning applications that I can reference?
I would consider a different platform, the Arduino is limited in memory and the needed processing power for image applications. I am not saying it could not do it but expect an extremely slow system.
Keep in mind that Arduino != AVR. There are Arduino boards that have quite a bit of memory and processing power. Machine learning doesn't take so much as you might thing. Keep in mind that you train the model on a PC. If you spend 30 seconds on Google you'll find plenty of Arduino-based machine learning projects working with image data.
I understand what you are saying and basically you said the same thing by using the PC for the training, I put it in the Arduino. My experience is in manufacturing automation where part checking etc at high speeds.
A very few do, but it is not worth the effort to start with any of them.
Learn the ML basics first, using your laptop or desktop PC. There are plenty of free tutorials, examples with data and software on line, and memory and disk space are usually not a problem. The latter will ALWAYS be issues with microcontrollers.