Week 3 + Week 4

[farzaa]

Jonah my guy.

So we are continuing the vision stuff. I just told you about the Google Vision API today, and you're going to sue that for OCR. So go ahead and set that script up.

There is also a lot of set up to do with actually saving everything that the API throws out. So here's what I'd do.

1. Set up script to read a melee video file, save the frames into images on your PC.
2. Set up script to take those images, crop just the percents, and send them to Google OCR API
3. Once you get results back the Google OCR API, you wanna save them because making requests there are expensive. So, take the results and save them in a JSON where every single frame file name has percents associated with it. I'd set it up like below

[{"file_name": "frame_54.jpg", "percent_1": "110%", "percent_2": "100%"},
{"file_name": "frame_55.jpg", "percent_1": "120%", "percent_2": "101%"},
{"file_name": "frame_56.jpg", "percent_1": "130%", "percent_2": "102%"},
{"file_name": "frame_57.jpg", "percent_1": "133%", "percent_2": "103%"}]