PAVIS is the Pattern Analysis and Computer Vision research line at IIT. PAVIS research is focused on artificial intelligence, computer vision, machine learning.
Genova, Italy
Joined June 2010
- Tweets623
- Following131
- Followers718
- Likes1.5K
PAVIS IIT retweeted
Amazing news, I will start the new year by joining Generative Bionics as Chief AI Officer.
After closing the round, now we can start to build autonomy for our humanoid robot.
Thanks to our investors and my co-founders for the incredible support during this year!
Generative Bionics announces a €70M #funding round.
Our mission is amplifying human potential by developing intelligent robot companions based on Physical AI.
👉 Learn more: gbionics.ai/
Stay tuned: we’ll be unveiling our first humanoid robot concept at @CES.
PAVIS IIT retweeted
Don't miss our morning poster today at @ICCVConference:
10:45-12:45 Poster Session 5
Highlight Paper
428 - Embodied Image Captioning: Self-supervised Learning Agents for Spatially Coherent Image Descriptions, T. Galliena, T. Apicella, S. Rosa, P.Morerio, A. Del Bue, L. Natale
PAVIS IIT retweeted
Going to @ICCVConference! @PAVIS_IIT has three paper in the main conference, check the thread. See you at the conference!
Today Matteo Bortolon defending his PhD work done in collaboration with @TeV_FBK, PAVIS and @UniTrento on Neural Representations for Robotics. Congratulations Doctor! 🎓 – at Trento, Trentino-South Tyrol
Congratulations to @alessiodelbue for this appointment at @durham_uni
@alessiodelbue, Principal Investigator of @PAVIS_IIT unit, has been appointed Honorary Professor in the Department of Computer Science at @durham_uni, in recognition of his pioneering contributions to computer vision, artificial intelligence, and their real-world applications.
PAVIS IIT retweeted
🎓 DCitizens was in Genova last week for a 2-day training at IIT’s Center for Human Technologies!
We explored research funding, ethics & Horizon Europe, and deepened collaboration between teams from @iti_larsys, @istecnico & @IITalk.
💡Real impact starts with strong foundations.
PAVIS IIT retweeted
Great to be back @IITalk and @PAVIS_IIT talking as part of @dcitizens_proj. Thanks for the invitation, lots of interesting chats.
@durham_uni @comp_sci_durham
PAVIS IIT retweeted
Well done Matteo Toso and Alessio Del Bue (@alessiodelbue) for presenting our work “Maps from Motion (MfM): Generating 2D Semantic Maps from Sparse Multi-view Images” at @3DVconf
Paper at: stuart-james.com/publication…
@IITalk @comp_sci_durham @durham_uni
Congratulations to our new Doctor Dario Serez!!! He successfully presented at the final year Ph.D. exam his work titled: "Multiple Latent Variable Generative Models: an Information-Theoretic Perspective and Applications"
Well done!!!
PAVIS IIT retweeted
Great collaboration between Naver Labs and @PAVIS_IIT! Check the website for our #CVPR2025 paper:
europe.naverlabs.com/researc…
@IITalk – at Camden Town, London
Our e2d trained navigation agent has been accepted to #CVPR2025: fast-moving, efficient: correct modelling of motion during training in simulation.
arxiv.org/abs/2503.08306
By S Janny, H Poirier, L Antsfeld, G Bono, G Monaci, B Chidlovskii, F Giuliari, A Del Bue and myself.
A group dinner eating pizza at PAVIS before the last rush for @ICCVConference #ICCV2025 deadline!
Even if the conference is in Hawaii this year, pineapple on pizza is forbidden in Italy! 🍍🚫
PAVIS IIT retweeted
MULA 2025, 8th Multimodal Learning and Applications Workshop mula-workshop.github.io/ CVPR co-located is accepting submissions!
Deadline is March 12th
Share your work with us on our workshop became a don't-miss event in our community! @paolorotaphd @pmorerio @michaelyangbath
...and today we have @efrontoni from the VRAI group of University of Macerata with his talk on the challenges of pushing the boundaries of computer vision research while ensuring successful enterprise adoption.
PAVIS IIT retweeted
⚡𝐄𝐱𝐩𝐥𝐨𝐫𝐞 𝐡𝐨𝐰 𝐀𝐈 𝐞𝐦𝐩𝐨𝐰𝐞𝐫𝐬 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 𝐰𝐢𝐭𝐡 𝐃𝐫. 𝐃𝐞𝐥 𝐁𝐮𝐞 𝐢𝐧 𝐨𝐮𝐫 𝐋𝐈𝐕𝐄 𝐬𝐞𝐦𝐢𝐧𝐚𝐫: youtube.com/live/gZXoek0_Ipw
🗓️ Monday, Dec 16th at 10:00 AM (GMT+7)
🔎Topic: 𝐅𝐫𝐨𝐦 𝐒𝐩𝐚𝐭𝐢𝐚𝐥 𝐀𝐈 𝐭𝐨 𝐄𝐦𝐛𝐨𝐝𝐢𝐞𝐝 𝐀𝐈: 𝐓𝐡𝐞 𝐏𝐚𝐭𝐡 𝐭𝐨 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐒𝐲𝐬𝐭𝐞𝐦𝐬
🏆𝐌𝐞𝐞𝐭 𝐭𝐡𝐞 𝐒𝐩𝐞𝐚𝐤𝐞𝐫 Alessio Del Bue is the director of the PAVIS (Pattern Analysis and computer VISion) research unit of the Italian Institute of Technology (IIT) in Genova, Italy. Previously, he was a researcher in the Institute for Systems and Robotics at the Instituto Superior Técnico (IST) in Lisbon, Portugal. Before that, he obtained my Ph.D. under the supervision of Dr. Lourdes Agapito in the Department of Computer Science at Queen Mary University of London. His current research interests are related to 3D scene understanding from multi-modal input (images, depth, audio) to support the development of assistive Artificial Intelligence systems. He is co-author of more than 150 scientific publications, in refereed journals and international conferences, member of the technical committees of important computer vision conferences (CVPR, ICCV, ECCV, etc.), and he serves as an associate editor of Transaction of Image Processing (TIP) journal. Finally, Dr. Del Bue is an IEEE and ELLIS member in the recently formed Genoa unit.
📘 𝐃𝐢𝐬𝐜𝐮𝐬𝐬𝐢𝐨𝐧 :
This talk explores our journey from Spatial AI, focused on environmental perception, to Embodied AI, where systems interact and learn from their surroundings. Leveraging advanced visual scene representation techniques like NeRF and 3DGS, we can create realistic, semantically rich models to train autonomy tasks such as navigation and object detection. Finally, we demonstrate how visual and multi-modal data enable Embodied AI models to power humanoid robotic platforms at IIT.
PAVIS IIT retweeted
Today, we present a new real-world archaeology dataset for puzzle solving in frescos at #NeurIPS2024 as part of the @repairproject21.
Join us from 16:30—19:30, West Ballroom A-D, Poster #5210
@IITalk @bengurionu @CaFoscari @unil @LaStatale @pompeii_sites @comp_sci_durham @unige – at Vancouver, British Columbia
PAVIS IIT retweeted
Code dropped: github.com/david-svitov/BBSp…
BillBoard Splatting (BBSplat): Learnable Textured Primitives for Novel View Synthesis
Contributions:
• We propose BBSplat with optimizable textured primitives to learn 3D scene representation with photometric losses. BBSplat allows up to ×2 acceleration for NVS compared to the state-of-the-art for the same rendering quality.
• We developed an algorithm to efficiently represent and store textures for billboards. BBSplat demonstrates significant storage cost reduction. As a result, for some scenarios, storing the scene represented with billboards is more efficient than 3DGS or 2DGS.
• We implemented BBSplat with CUDA and demonstrated its efficiency in extensive experiments on several real-scene open datasets.
