Last year we added CLIP-based image search to https://immich.app/ and even though I have a pretty good understanding of how it works, it still blows my mind damn near every day. It's the closest thing to magic I've ever seen.
Happy immich user here! I once took a cute photo of our baby
chewing on a whisk, and actually finding the correct photo in an unsorted, untagged huge pile of photos by simply searching for "whisk" was a mindblow experience! It is an amazingly powerful tool!
How does it compare to Google Photo search? I search things like 'whisk' with success regularly... though to be fair not as random as whisk, but things like "steering wheel"
I'd also consider adding searching via QR codes. you could search by the content in the QR code (like the url) or if its a URL, search the content on the page of the QR code.
people that have a lot of photos with QR codes would want it :).
You could search your images for qr codes that go to linkedin, ig, or fb pages. or find the qr code wifi passwords.
if you just have a screenshot of a QR code (like you zoomed in and screenshot the ticket's QR code, no other text), then finding the qr code by the event name could be useful.
Thailand requires qr codes that linked to the nutritional information registered online. this could be useful to help you search products that just show the back of the label and not the front of the product.
For anyone in the same boat as my self, I later found out that this is actually very easy to achieve (thanks to ChatGPT). Theoretically, this is how it is done
1. Encode faces, there is a library called face_recognition, that can grab faces from pictures and encode them
2. Group the faces data using `pairwise_distances(encodings, metric='euclidean')`, you only need sklearn library for this
