HackStack

Project Info

VibeS

Devpost

Inspiration

While at the CalHacks venue, a team member, Jeffrey, took a picture of the city buildings and wondered, "How can we capture the vibe of this picture?". The team had previously wanted to work on an AI and music-related project, so we got to work in trying to successfully capture the VibeS.

What it does

Provided a picture and a Spotify account, the application will provide a curated playlist that captures the VibeS of the picture based on the user's preferences.

How we built it

We trained a visual transformer BEIT from Hugging Face using Intel's Cloud Computing services that categorizes pictures. Using some of these categories, we then fetch from a Convex database populated with the user's songs from his "niche" playlists and provide the top results.

Challenges we ran into

Making a dataset for the visual transformer was challenging as well as learning new technologies like Convex database mutation and querying.

Accomplishments we're proud of

We are proud of building accurate models for the time of the day and environment classifications. We are also proud of being able to build an appealing front end for our project.

What we learned

We learned a lot of skills such as training and testing models using Hugging Face and Intel's Cloud Computing Services, using Convex for database mutations and queries, fetching user information using Spotify's API, and getting more comfortable building in TypeScript and React.

What's next

More data, more classifications for images, and more features such as instantly adding a playlist to Spotify, linking results to Spotify song links, etc.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • HTMLIn code
  • JavaScriptIn code
  • ReactIn code
  • Tailwind CSSIn code
  • TypeScriptIn code
  • Hugging FaceClaimed

6 of 7 appear in the indexed code. 1 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

50 KB

Source files

34

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars