Inspiration
Have you ever looked through a messy bookshelf and couldn’t decide what to read? With ShelfMate your bookshelf is digitized and you can easily look for the book you want.
What it does
ShelfMate is a user-friendly website that allows users to effortlessly view the contents of their bookshelves by simply taking a photo of the books on the shelf.
How we built it
The PaddleOCR program then processes the image and extracts the title and author information from the book bindings. The API retrieves book titles, author information, ratings, page counts, and a brief description.
Challenges we ran into
After trying out different OCR methods and libraries, we found that PaddleOCR was the best option for our needs. However, the OCR-generated text is often jumbled and disorganized, especially when extracting text from multiple books placed close together. So, we decided to train an Assistant on the OpenAI API to clean up the text and make it suitable for input into the HardCover API.
Accomplishments that we're proud of
We are proud of the progress we were able to make in such a short amount of time. The website and backend is fully functional which is a great milestone.
What we learned
We learned about all the great technologies that made this project happen!
What's next for ShelfMate
For ShelfMate, we want to add more features in the future and maybe integrate an app as well.
Log in or sign up for Devpost to join the conversation.