Discova is a platform designed to improve the discoverability of NFTs. It utilizes computer vision-based image processing for a contextual understanding of the NFTs(image usecase implemented) followed by text-based vector & semantic embeddings and storage on a Vector Database(Qdrant implementation - can be easily switched to pinecone or any other VectorDB).
The infrastructure was designed keeping in mind the development down the line - utilizing a Python-based framework for leveraging state-of-the-Art computation Libraries, whereas the frontend was developed in NextJS for inter-operability on the blockchain. The solution utilizes several web3 technologies and provides a semi-decentralized approach that can be ultimately fully decentralized using services like BitTensor.
This solution was developed as a hackathon project, and I am not planning to continue maintaining it.
If you're interested and want to collaborate for further development, please reach out on rutvikrj26@gmail.com or thomas.melbin4@gmail.com