Couldn’t be more grateful for this #ICCV2025 collaboration! It turns out that some vision models have actually been encoding images’ processing and acquisition parameters (e.g., JPEG compression settings, camera model) this whole time For more info check out @stojnicv.xyz‘s thread!

Feel free to visit IS1-081 if you’re at MIRU! An extended version was accepted at an ICCV workshop! Here we investigate whether bias in pre-trained CLIPs transfer to downstream tasks when you train models like LLaVA Co-authored w/ Yusuke Hirota, Yuta Nakashima, & @noagarciad.bsky.social

This is one of those situations where everything he said applies to me too lol I have a super caring genius of an advisor in @noagarciad.bsky.social. She does so much for us, hard to put into words the amount of gratitude she deserves Hoping to pay it all back in 2025 and onwards