SwingSearch

• 

  Landing Page

Table 4

Inspiration

As a swing dancer who dances around the country regularly, I have noticed it's hard to DJ for swing dancing since someone has to sit there to 'vibe check' the room - so we decided to make a camera vision project that monitors the mood in a room and automatically DJs your swing dances for you! This way nobody has to be left out and everyone can join in on the fun. subtle plug

What it does

• Uses audio analysis to create a cosine similarity matrix to compare songs
• Use audio analysis to estimate the BPM of each song
• Uses computer vision to track user movement and speed to get the vibe of the room
• Uses a recommender system to recommend songs that are similar or completely different to the current song based on preference

How we built it

• using Python, Webflow, Librosa, Figma and Open CV
• Scraped youtube for swing songs
• Use scipy and librosa to estimate the BPM of each song as well as similarities in the frequency domain.
• Used figma to generate a design for the frontend and webflow to turn that into code
• Had a swing dance lesson to understand the cultural significance of swing as well as gain the skills for the demo
• Use OPENCV with the YOLOv3 model and Simple Online Realtime Tracking (SORT)

Challenges we ran into

• Spotify API not being straightforward
• Webflow issues
• Computer Vision errors with MacOS

- Most of the team are first-time hackers

Accomplishments that we're proud of

• Used computer vision for the first time
• Completed demos of frontend and backend

What we learned

• How to use Computer Vision
• Using Librosas library

What's next for SwingSearch

• Integrate Spotify and the frontend (which is pretty btw!)

Built With

• python