Quote (lilpsychokilla @ Sep 7 2019 08:03pm)
This is not an easy beginner project. You should first get a good grasp basic code writing for python. If/else statements, functions, classes, operators, indentation etc. This specifically is going to be math heavy. You'll need a database of all the possible conversions. Each compoud is going to have to be associated with several number values for each shape. The drawn version will be converted to a number value that will be matched with a compound number value. I don't know much about what compounds look like but id assume you'd need a value for each line. I'd recommend you looking up some premade image recognition code on github. You could probably save a lot of time just by using stuff already out there and tweaking it to your needs, but first learn the basics of python and coding in general. You'll definitely need to learn how to use pip as well. This website is pretty dogshit for coding, find some youtube tutorials, get familiar with python documentation, and defintely solve algorithms daily either through a website or preferably a book. Stackoverflow is your best friend, ask questions there, not here. I've never made anything that does image recognition but I'm pretty sure you will need a good understanding of matrices and manipulating arrays in arrays, which is around intermediate difficulty.
You'll probably need a couple hundred hours in learning before you can even get started to be honest.
This. Only going to add one thing:
There is a free trial of wikitude that may be able to save you some time without actually having to do anything other than train the images.
Learn the fundamentals and check out Google Cloud Platform, Vuforia, Wikitude, and look at some openCV projects. Stackoverflow+YouTube will get you further in a day than jsp will in a year lol.
If you decide to do it let me know
This post was edited by Sn0 on Sep 8 2019 10:30pm