Benedetta Liberatori

What does it really mean for two videos to be similar? At #NeurIPS2025, we’ll present ConViS-Bench: Estimating Video Similarity Through Semantic Concepts. Stop by our poster for a chat! 📍Exhibit Hall C,D,E — Poster N.4618 🕒 Thu, Dec 4, 2025 • 11:00 AM – 2:00 PM PST

@valeoai Thanks for having me!

Our visiting PhD student @bliberatori_ gave us a great talk on Video understanding beyond trimmed clips: temporal and comparative reasoning. Make sure to have a look at her excellent works in this area: benedettaliberatori.github.io

1/n New paper - V-GIFT 🎁 Self-supervised tasks like rotation prediction or colorization were big in 2018. Do they still matter? Yes. We turn them into visual instruction tuning data for MLLMs. Result: models rely more on the image and perform better on vision tasks 👀

Huge congrats to our colleague and friend @lucazanella95 on his graduation! 🎓 Brilliant researcher, great teammate, and a wonderful person, we’re proud to celebrate this milestone with you, Luca! Looking forward to what comes next 💪

🚨 Are we underestimating Large Multimodal Models for classification? We challenged the status quo in our new paper from @mhug_unitn, "Large Multimodal Models as General In-Context Classifiers", accepted at #CVPR2026 Findings 🔥 A quick thread on our discoveries 🧵👇1/7

We are so back 😎

Can MLLMs ask for help? 🤔 When facing an unanswerable visual query, models typically hallucinate or abstain. We argue there's a better option: being proactive and requesting a simple user intervention. Introducing ProactiveBench! 🧵

Just in case you miss us! DEMO is now in #3DV2026 poster session 3, N.20! We propose a new task: Dense Motion Captioning, together with the dataset: CompMo (60k, ~40s duration) and the model: DEMO for the task 🕺🏽project page: xusy2333.com/demo/ come and say hi (virtually!)

working on human motion and in #3DV206 ? then don’t miss this interesting work! in “Dense Motion Captioning”, the authors propose a new task, dataset, and the model for dense 3D human motion captioning

The deadline is coming… and it waits for no one. 🐺 📚 MULA workshop submissions close tomorrow (March 9) Details: mula-workshop.github.io #CVPR2026

🎉 Big news! 🎉 We’re excited to announce that the 9th edition of the Multimodal Learning and Application Workshop (MULA) will be held at CVPR 2026 in Denver as a full-day workshop! #CVPR2026

I'm happy to announce that our latest paper, “Group-Robust Machine Unlearning,” has been accepted at TMLR! 📄 arxiv.org/pdf/2503.09330

Check out the project page to see more: lnkd.in/dqejbRhe Huge thanks to my coauthors: @altndrr , Lorenzo Vaquero Otal, @YimingWang107 , @eliricci_ , @paolorotaphd.

What does it really mean for two videos to be similar? At #NeurIPS2025, we’ll present ConViS-Bench: Estimating Video Similarity Through Semantic Concepts. Stop by our poster for a chat! 📍Exhibit Hall C,D,E — Poster N.4618 🕒 Thu, Dec 4, 2025 • 11:00 AM – 2:00 PM PST

📢 A quick broadcast during the rushing hour of CVPR chaos! 👯‍♂️ Our work: Dense Motion Captioning got accepted to #3DV2026 !!!🎉 TL;DR: We propose the dense motion captioning task and the first dataset combined with a simple model. 🕺🏽 Details and visual results, please see the project page: xusy2333.com/demo/ 🥺 Huge thanks to my advisors and coauthors @bliberatori_ @gulvarol @paolorotaphd 📄 Paper link: arxiv.org/abs/2511.05369 🤔 Also, a question I keep pondering: Is VQ-VAE really needed as the human motion tokenizer for understanding? ...Well, I don't think so. What's your opinion?👀

Dublin, here we come! 🇮🇪 Our paper "Dynamic Scoring with Enhanced Semantics for Training-Free Human-Object Interaction Detection" has been accepted to #ACMMM2025

Had a great time yesterday afternoon presenting our poster on Multi-focal Conditioned Image Synthesis with Nicu! Huge credit to Jiaqi, our first author, whose excellent work made it all possible, but could not come to #CVPR2025 👏🧠🖼️

🚀 Excited to share that our latest work "Unlearning Personal Data from a Single Image" is published at TMLR! 🌟 A big thanks to my co-authors: Massimiliano Mancini, @Steph_lat, @sroy907, @eliricci_.

🚨 CVPR main conference decisions are out, unlucky? Do not despair! The MULA workshop deadline is March 12, 2025. If you want to experience Nashville and present your work, don’t miss this opportunity! 🎸📅 #CVPR2025 More info here 👇🏻 mula-workshop.github.io

How's your @CVPR submission? Us:

🤖🔍 Want to continually learn vision-language tasks without relying on external data? Check out our latest approach, GaB, now accepted at #WACV2025! It tackles catastrophic forgetting head-on. Learn more: arxiv.org/abs/2411.02210 #AI #MachineLearning