So, I’m selfhosting immich, the issue is we tend to take a lot of pictures of the same scene/thing to later pick the best, and well, we can have 5~10 photos which are basically duplicates but not quite.
Some duplicate finding programs put those images at 95% or more similarity.
I’m wondering if there’s any way, probably at file system level, for the same images to be compressed together.
Maybe deduplication?
Have any of you guys handled a similar situation?
Definitely PhD.
It’s very much an ongoing and under explored area of the field.
One of the biggest machine learning conferences is actually hosting a workshop on the relationship between compression and machine learning (because it’s very deep). https://neurips.cc/virtual/2024/workshop/84753
The problem is that OP is asking for something to automatically make decisions for him. Computers don’t make decisions, they follow instructions.
If you have 10 similar images and want a script to delete 9 you don’t want, then how would it know what to delete and keep?
If it doesn’t matter, or if you’ve already chosen the one out of the set you want, just go delete the rest. Easy.
As far as identifying similar images, this is high school level programming at best with a CV model. You just run a pass through something with Yolo or whatever and have it output similarities in confidence of a set of images. The problem is you need a source image to compare it to. If you’re running through thousands of files comprising dozens or hundreds of sets of similar images, you need a source for comparison.
computers make decisions all the time. For example, how to route my packets from my instance to your instance. Classification functions are well understood in computer science in general, and, while stochastic, can be constructed to be arbitrarily precise.
https://en.wikipedia.org/wiki/Probably_approximately_correct_learning?wprov=sfla1
Human facial detection has been at 99% accuracy since the 90s and OPs task I’d likely a lot easier since we can exploit time and location proximity data and know in advance that 10 pictures taken of Alice or Bob at one single party are probably a lot less variant than 10 pictures taken in different contexts over many years.
What OP is asking to do isn’t at all impossible-- I’m just not sure you’ll save any money on power and GPU time compared to buying another HDD.
Everything you just described is instruction. Everything from an input path and desired result can be tracked and followed to a conclusory instruction. That is not decision making.
Again. Computers do not make decisions.
The first thing I would do writing such a paper would be to test current compression algorithms by create a collage of the similar images and see how that compares to the size of the indiviual images.
Compressed length is already known to be a powerful metric for classification tasks, but requires polynomial time to do the classification. As much as I hate to admit it, you’re better off using a neural network because they work in linear time, or figuring out how to apply the kernel trick to the metric outlined in this paper.
a formal paper on using compression length as a measure of similarity: https://arxiv.org/pdf/cs/0111054
a blog post on this topic, applied to image classification:
I was not talking about classification. What I was talking about was a simple probe at how well a collage of similar images compares in compressed size to the images individually. The hypothesis is that a compression codec would compress images with similar colordistribution in a spritesheet better than if it encode each image individually. I don’t know, the savings might be neglible, but I’d assume that there was something to gain at least for some compression codecs. I doubt doing deduplication post compression has much to gain.
I think you’re overthinking the classification task. These images are very similar and I think comparing the color distribution would be adequate. It would of course be interesting to compare the different methods :)
