get last features for similarity

[rastinrastinii]

Hi

I'm interested in using your virtual try-on repository ([repository name]) for a project involving image similarity. I'd like to leverage the model's ability to extract features from images (clothing items) to compare them and find similar styles.

Specifically, I'm looking for guidance on the following:

Feature Extraction:
Could you point me to the part of the code responsible for extracting the final features from the clothing images in the virtual try-on process? Are there any specific pre-trained models or layers used for feature extraction?
If the features aren't readily available, would you be able to suggest modifications to extract them at a suitable point in the pipeline?

Image Similarity:
Once I have the features, what approach would you recommend for comparing them to find similar clothing items? Are there any specific distance metrics or similarity measures that work well in this context?
Any considerations or potential challenges I should be aware of regarding image similarity within the virtual try-on domain (e.g., pose variations, lighting differences)?
I've looked through the repository documentation, but I'd appreciate any additional insights or code snippets that could help me achieve this.

Thanks in advance for your time and assistance!