Show HN: I made a Pinterest clone using SigLIP image embeddings

Sometimes there are duplicate results, e.g. <a href="https:&#x2F;&#x2F;mood-amber.vercel.app&#x2F;images&#x2F;0b733fc2-7093-4443-8872-016961c54000" rel="nofollow">https:&#x2F;&#x2F;mood-amber.vercel.app&#x2F;images&#x2F;0b733fc2-7093-4443-8872...</a> has two copies of <a href="https:&#x2F;&#x2F;mood-amber.vercel.app&#x2F;images&#x2F;f920a599-bbd7-4805-3317-4a8531fb5800" rel="nofollow">https:&#x2F;&#x2F;mood-amber.vercel.app&#x2F;images&#x2F;f920a599-bbd7-4805-3317...</a> right next to each other. (The link UUID is the same, so I assume this is an issue with the search algorithm, not simply duplicate data that got scraped.)

Also, check <a href="https:&#x2F;&#x2F;same.energy&#x2F;" rel="nofollow">https:&#x2F;&#x2F;same.energy&#x2F;</a>

Very cool! How did you get the collage layout? I noticed images in each column don’t have the same size. I assume images get Centre cropped?

Cool! I haven’t tried SigLIP out yet but it seems to be the new hotness over CLIP… I just dont have a good project idea yet

Is there a repo, especially for training? I&#x27;d like to see how SigLIP performs on a dataset of only anime images.

There are some interesting images there. Why are you not including the source of the images?

Good stuff! Do you have any intuitive sense of whether SigLIP is particularly stronger than CLIP here? Also vector DB over Faiss index?

Cool project! Thanks for sharing

Neat! How many images are in the dataset out of curiosity?

how far we&#x27;ve come since <a href="https:&#x2F;&#x2F;www.karlsims.com&#x2F;genetic-images.html" rel="nofollow">https:&#x2F;&#x2F;www.karlsims.com&#x2F;genetic-images.html</a><p>quite a bit, but surprisingly not

nice, we always need more clones and improvements.... hope you get traction.<p>I never click Pinterest links because the experience is too bad.