Dynamic Creative Optimization for AdTech
Platform: Web (Cloud-based AI)
This project analyzes the technical feasibility of applying generative artificial intelligence technologies in the AdTech field through a proof of concept, focused on automating visual content creation for advertisements. *Disclaimer*: I've participated in this project *only* on my role of tutor.
The creation of visual content for online advertising is often a manual and expensive process. This project explores the use of generative AI to automate this workflow, enabling the quick generation of high-quality product assets.
A significant challenge identified was the generative models' lack of prior information regarding new products. To overcome this, the team employed fine-tuning techniques, notably Dreambooth with Stable Diffusion, allowing the models to learn from specific product images and data. This approach significantly improved the precision and relevance of the generated visuals.
To ensure the solution is both scalable and cost-effective, a cloud architecture was implemented using Amazon Web Services (AWS). The result is an efficient infrastructure that maintains high quality without high costs. The project concludes that developing such a system to streamline creative production is viable, provided the discussed implementation challenges are addressed.
Authors
Disiot Mendizabal, Agustina | Monjardin López, Iván Rodrigo | Diaz Hugo, Santiago
Reviewers
Nieves Lema, Ruben Carlos | Matalonga Motta, Santiago
Score
100/100
