What it does
Vido takes your long video and transforms it into easily consumable content. It uses a dynamic programming algorithm to determine which parts of a video have the most relevance. It highlights those parts and cuts out the rest so you get a video that addresses the parts of most concern.
How we built it
Vido is built upon a chain of data translations. We start with a source video file, extract the audio, and then pass that through Google's speech-to-text. We map those words to sentences and pass those through a text summarizer specially built to return video. Finally, we use the timestamps provided by the summarizer to edit the old video which gets served up by the web app.
Challenges we ran into
Text parsing is difficult, and videos are not easy to work with. Reactjs integrated with our Flask server, but not easily.
Accomplishments that we're proud of
This project was a success in many ways. This is not an easy feat, and many of the technologies we used were new to all of us. The project worked much better than expected, and the summaries are more comprehensive and smooth. The web integration was also a success. Our team communicated well, and apart from some minor complications, we completed each of our goals with minimal git issues.
What we learned
All of the aspects of our project were new to the team member who worked on it. This project was a large departure from everything else we've done.
What's next for Project Vido
We'll be at HackCU VII.
Log in or sign up for Devpost to join the conversation.