Welcome to Boardsort™ - Learn - Sell - Profit -

im not scraping at the moment, doing other things. but ive been getting into chat-gpt lately and came across this forum again. im wondering what some big players think about using some sort of neural net, or even a custom gpt (even though its a language model and not rele the best for images), to help in assisting sorting. am i late to the game and this is already implemented? this can't be a new idea, is it? eventually when gpt's open up a bit more, this could also be implemented online and replace the "What is it? Identification Forum". am i late to the game, giving away an amazing idea, or maybe some unknown spot in between??

How will AI interpret a photo for precious metal content and classification?

Probably by spotting common components (IC chips, gold pin connectors, tantalum capacitors, etc.) in addition to considering things like areas of boards with mostly worthless components, heatsinks, transformers, etc.

This seems like something that can be done by AI, if someone had the time to develop one.

The hardest part would be making the data set. With 50+ different categories it would take at least several thousand photos. With 80 categories and 50 photos each (which isnt that much i dont think) thats 4000 photos. After that i think its actually quite straight forward. With that available I would be fine taking it on as a side project.