Hello, I am using the OpenMV in a project and came across a video on YouTube that talks about the char 74k: https://www.youtube.com/watch?v=MA884nplBYc. I would like to know how I could obtain the network file, and if it would be worth it, since I only need to identify the letters H, S, and U. Also, if using a network is not very good, what would you recommend for quickly and concisely identifying these letters?
Hi, that video is extremely old. If you want to do this now please use Edge Impulse to train a CNN that can do this. We can run any quanitzed TensorFlow lite CNN onboard. So, given this you just need to find a dataset you can do transfer learning on to create the model and then you can deploy it. If you ask for help support on their forums they may be able to point you to someone’s project who has already done this too.
Ok, thank you, but I have another question. Is there any other way to identify the letters H, S, U without using CNN?
Yes, use the find_line_segments() method. It will give you a list of line segments that you can try to match to each character.
Ok, I will try and bring the results tomorrow