[FR] Very cool, couple of additional ideas

[andrewginns]

This is great! I've been looking for something like this for quite a while. I have two suggestions:

Output results in yolo txt format similar to RectLabel:
https://rectlabel.com/

Allow for inference from a live video feed:
https://developer.apple.com/documentation/vision/recognizing_objects_in_live_capture
a. webcam for testing purposes
b. video capture card BMD decklink and ultrastudio: https://www.blackmagicdesign.com/products

Allow user to select how often inference is run on video, i.e., perhaps instead of running on all frames a user could select 1 frame per second, etc.

From: npna#1 (comment)

@lonnylundsten