I'm primarily working with video-intelligence on a dataset of videos for some academic research. I'm a complete novice, but I've started to get the hang of things, and I'm getting some really useful outputs from speech to text, text recognition, and labelling.
However, in the vision API, there is a 'detect image properties' function that looks really promising. If I could detect things like dominant colour for shots within my videos, and then get an an average output for each video as a whole, this could really help with my analysis.
The problem is, I don't know how much flexibility there is in this kind of thing, and I don't have enough fluency with this kind of work to even know how to ask (I only just about understand a for loop lol).
My current thinking is run the analysis on each shot of a video, attribute weights to outputs, and then take a weighted average of some sort? This is only pie in the sky thinking though. Any thoughts welcome, I'm just exploring right now.
LinkedIn
Twitter
