Reverse engineer video ads frame by frame using Vision API (this outperforms ChatGPT)

Nov '23 (edited) • 💬 Discussion

I got ChatGPT to write a Python script that lets you upload a video and then it:

1. Transcribes the audio

2. Extracts the image frames (at 1 fps)

3. Describes each image frame

4. Analyses both the audio transcript and the image frames

Nothing revolutionary but considering that I have 0 coding experience I was pretty happy with the end result 😛

You can find the Colab file here. You can run it after making a copy of it.

Hopefully some of you will find it helpful. Let me know what you think!

4 comments

skool.com/content-academy

Build a profitable content strategy, efficient team & automated process

Leaderboard (30-day)

+51

+48

+34

+23

+17