Here’s the revised version:
Inspiration
We wanted to find a way to make creative analysis easier to understand and, most importantly, to turn it into useful recommendations for users who need guidance, not just data.
What it does
Our web app combines computer vision and CSV campaign data to evaluate performance in an interpretive way, explain what is working, and suggest actionable recommendations.
How we built it
We built the project with a Python backend and a JavaScript web frontend, connecting the data pipeline, prediction logic, and user interface into one interactive workflow.
Challenges we ran into
One of the main challenges was making the visual resources match the written data, since they did not always align well. We also had no previous experience with computer vision, which made it harder to identify and extract certain variables reliably.
Accomplishments that we’re proud of
We are proud of building a system that uses computer vision to extract insights from creatives, combines them with campaign data, and turns them into understandable recommendations. We are also proud of implementing a conversational chat using an external API, which works as a coach for the user.
What we learned
We learned that good AI tools are not only about prediction, but also about clarity, explanation, and helping users make better decisions.
What’s next for Creative Intelligence
Next, we’d like to improve our model with better and richer data so it can make more accurate predictions and stronger recommendations.
Log in or sign up for Devpost to join the conversation.