Is there a model for image pair duplicate detection?

Beginners
1

Is there a model in hugging face for duplicate image pair detection?

I have a dataset with millions of image pair with labelling that they are duplicate or not:

image1, image2, duplicate
image3, image4, near-duplicate
image5, image1, not-duplicate
.........
.........
.........
image99999, image8383920, not duplicate

Is there a model in hugging face that I could use? During inference the model would produce the following verdict:

(imageX, imageY) = verdict?

verdict = duplicate or not duplicate or near duplicate.

1 Like
2

You can consider using image similarity estimation using a siamese network. A couple of examples that are worth looking at:

Siamese Networks for Duplicate Image detection

Image similarity estimation using a Siamese Network with a triplet loss

Duplicate Query Detection using Siamese Network

Siamese coding network and pair similarity prediction for near-duplicate image detection

Using Siamese Network for Duplicate Detection in MxNet

A Scalable Solution to Detect Duplicate Images

3

You have shared some very useful examples. I have learned a lot while exploring all of these examples specifically from “A Scalable Solution to Detect Duplicate Images” and “Using Siamese Network for Duplicate Detection in MxNet”.