Hey everyone! I'm a recent graduate diving into a project to sharpen my skills and enhance my resume. I'm creating a mobile app that uses AI/ML to detect issues in car tires based on user-submitted photos (I've got some experience from working at a tire shop for a year, so I thought this would be a great focus). Right now, I'm trying to figure out the best way to set up my application's database. I've heard that a vector database like Pinecone might be useful for this, but I'm uncertain if that's the right direction. Being new to both databases and AI, I want to ensure I'm not wasting time on the wrong approach. Is Pinecone suitable for storing images and tire damage data, or should I consider another option? Also, when should I start building the classifier for this app? For context, I'm planning to write the app in Java since I want to learn that language. Thanks in advance for your advice!
1 Answer
Before you settle on a database, think about what exactly you'll be storing. Images can be tricky because they're not very database-friendly. If you're aiming to do any vector searching – which Pinecone is great for – you need to have something to vectorize. Otherwise, it might be better to look into a traditional database for image storage and relevant metadata.
I plan to store images along with details like tire damage types, DOT numbers, and manufacturers. I think identifying damage is straightforward, so I won't need tons of images for submissions. I wonder if there's a way to relate issues through vector searches instead?