Inspiration
Looking at the raw building blocks that make up a videofile, we determined a solid foundation for our model.
What it does
Transforms video-data into metadata
How we built it
Using advanced scrapers and a complicated algorithm using google cloud vision api, we determine keywords of important trailer-frames with the help of a ML-model. This dataset can be trained for metadata-correlation.
Challenges we ran into
Reliability of the dataset to train our models.
Accomplishments that we're proud of
Relatively high succes rate, in regards to de scarse dataset.
What we learned
Overfitting is a silent killer.
What's next for Case4 Team1
Lots of sleep and maybe a cool event-searching-app?
Log in or sign up for Devpost to join the conversation.