Project Overview

1. Explore the challenge data to understand various aspects of adverts like preview, landing frame, ending frame, ER, CTR. Perform EDA to get the highest performing adverts.
2. Image & text analysis using CV models to get the positions of assets within each frame. Get the color composition of assets and extract text from images.
3. Image composition using Pillow by defining their optimal positions.
4. Use Autogen for autonomous image composition.
5. Use GPT-4o to review composed ad frames and rate them out of 10 with respect to design principles.