The aim of this project was to build a multimodal model, accepting images and text, that predicts whether a video appears in the Trending category on Youtube.
You use this command to app run this locally with docker-compose: docker-compose -p thumbait -f azure.yml up
azure.yml uses images shared on Docker Hub:
- docker pull thumbait_main-server pszelew/thumbait-backend:latest
- docker pull pszelew/thumbait-frontend:latest
Models trained by us are available on gdrive: Models