📅 Join us next week for the 2nd VAR Workshop at CVPR 2026: June 3rd, 2026 from 8:30am to 1pm. 🎯 We are hosting an exciting line-line of speakers: Katerina Fragkiadaki, Wenhu Chen, Michael S. Ryoo, Ziwei Liu, Yao Qin, @vicenteor.bsky.social 👉Schedule/Papers: varworkshop.github.io/schedule/

Someone here needs to hear this today. Remember to rotate your tires and do alignment once in a while.

⏰The ICCV 2025 discussion phase will close soon! ⏰ As a reviewer, it's your job to: -📖 Read author rebuttals -🗣️ Engage in discussions -✅ Submit your final rating & justification Your responsible engagement is critical for a fair review process! Deadline: May 27!

My solidarity to Harvard colleagues and my respect for maintaining dignity during troubled times. Their leadership makes their standing in the global stage well deserved.

Cappuccino served at the 50th year celebration of the School of Engineering and Computing at Rice.

Group picture with my PhD students #studio-ghibli

Adding here an example of what I was going for: entrepreneurs4nsf.formstack.com/forms/petiti...

I remember when Twitter was this small and familiar. I had/have some followers on Twitter that now are low key celebrities but probably followed me long way back before they became low key celebrities.

#CVPR2026@cvprconference.bsky.social · last yr.

By popular demand, we are extending #CVPR2025 coverage to Bluesky. Stay tuned!

By popular demand, we are extending #CVPR2025 coverage to Bluesky. Stay tuned!

There are so many more things to be outraged about at the moment than complaining about federal funding for science. Especially when coming from academics, it is a bit too self-serving at this moment. Yes it is bad and the effects will be long lasting but so will be a dozen other things.

I think it’s exciting to see LLMs that are open source and on par with the top models accesible only through APIs. DeepSeek and before that Llama-3.

Can pretrained diffusion models be connected for cross-modal generation? 📢 Introducing AV-Link ♾️ Bridging unimodal diffusion models in one self-contained framework to enable: 📽️ ➡️ 🔊 Video-to-Audio generation. 🔊 ➡️ 📽️ Audio-to-Video generation. 🌐: snap-research.github.io/AVLink/ ⤵️ Results

Check this recent work by my PhD student Moayed. He has been doing amazing work on Generative AI for images, video and audio. We introduce AV-Link ♾️, an unified approach for audio-video generation. Our generated audio is the best in terms of synchronization with video actions. Check thread below.

Moayed Haji ALi@moayedha.bsky.social · 2y ago

Can pretrained diffusion models be connected for cross-modal generation? 📢 Introducing AV-Link ♾️ Bridging unimodal diffusion models in one self-contained framework to enable: 📽️ ➡️ 🔊 Video-to-Audio generation. 🔊 ➡️ 📽️ Audio-to-Video generation. 🌐: snap-research.github.io/AVLink/ ⤵️ Results

I still don’t feel quite more productive in the era of LLMs. There are very few things I can do better but far from what I hear from anecdotes. I wonder what would be the one low hanging fruit I should be delegating to LLMs.

Happy New Year Everyone!

End of year celebration with the great Moshe Vardi @myvardi.bsky.social is one of the perks of Rice. Hopefully he will be active on Bluesky soon.

Job alert! 🚨 (a bit special but anyway) If you are (or know someone who is) a Phd about to graduate or just graduated, and have to skip some in between time, I currently have a PostDoc position that could run for 6 months. Needed: Experience with publishing at Top A conferences. Just ping me. 👍

Had a great time this week visiting the University of Pittsburgh. I had not been to Pittsburgh in 10 years. I hope to be back sooner next time.

Looks like a fresh start in a new platform does not correct pile on behavior. It is hard to even notice when this is happening especially on one end. We see our own comments but we can’t see the other hundred comments saying the same thing in less polite language.

Lots of people talking about ICLR reviews. For me the most frustrating these days is when a reviewer declares that a paper has not reached the required bar for acceptance or novelty. As if this was an objective commonly agreed tenant that everyone knows.

La Era de la Inteligencia Artificial a short documentary produced by Telemundo Houston won a Lone Star Emmy in the Science category www.telemundohouston.com/noticias/tec...

La Era de la Inteligencia Artificial

Nos adentramos a la revolución de la inteligencia artificial con la historia detrás de Ameca, el robot humanoide más avanzado del mundo

telemundohouston.com

The last ten years in computer vision and multimodal AI have been a ride, yet I still feel there's so much more to explore.

Started a list of some researchers working on image/video generation. (Not comprehensive at all) Reply with a paper link and TLDR to get added to the list! I request all grad students to not feel imposter-y and just reply if you work in this field! #computervision #diffusion go.bsky.app/SP1uWoE

My growing list of #computervision researchers on Bsky. Missed you? Let me know. go.bsky.app/M7HGC3Y

Common software development advice that says we should write tests before implementations also applies to LLM-generated code. PropTest: Automatic Property Testing for Improved Visual Programming. arxiv.org/pdf/2403.16921

Rice’s Computer Science. View outside my office. Not judging just sharing.

Building a starter pack in computer vision. I will be adding more people as I encounter them. go.bsky.app/VNSuLYq

I never announced on Twitter that I would be leaving and I never really closed my account. Regardless of what is going on, and why. I believe it is healthy that we have several platforms and not just the one. Let’s see how Bluesky goes, it seems to replicate the old Twitter experience the best.

This year I was invited to attend the Dreamforce event in San Francisco — Salesforce’s biggest yearly event. I met Silvio Savarese and friend Juan Carlos Niebles. Amazing how large industry events are. Research conferences even in ML and CV don’t compare.

Let’s not bring Twitter grudges to Bluesky let’s start with a clean slate here. Thank you.