The challenge I’m having is mainly doing the detection on the numbers to start. Any help would be greatly appreciated.
Project Type:
I’m trying to create a project to detect the numbers within a paint by numbers image. And if possible in future use this information to build an application that could clean up the image and allow the user to select the numbers and highlight the various areas based on the number being detected.
Operating System & Browser:
I’m using Chrome, and Windows.
Hey Johan_Germishuys! Looks like your included link might be going to a private workspace. You could either make that project public or just add a screenshot of a sample image to help others troubleshoot here. But as a first guess, you might be looking to do object detection first to find “things that look like numbers” and then run OCR on that cropped out detection to get the actual number. But certainly there could be some more good options depending on what the data look like!
That helps for sure! (There is a method in computer vision called blob detection which might be able to isolate each section, at which point you can OCR for the number. I’m not sure to what degree you could run this on the Roboflow platform. Maybe others will have suggestions.)