📢 New Paper Alert! 🚀📄 Excited to announce "LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding." 📝Paper: lnkd.in/gHYG3ruX 🧑🏻‍💻Code: lnkd.in/gmxunM2C 🚀Project (Demo): lnkd.in/gqKVq25U

How can innovation protect our planet and expand human potential? The latest #HCI2025 Perspectives from KAUST explores how AI is helping communities thrive in harmony with nature. #BeyondReadiness #SaudiVision2030

Vlad Kim of Adobe sharing thoughts on interactive generative sculpting C3DV @CVPR25

3d compat challenge by Mahmoud Ahmed (kaust ) C3DV workshop

Prof Hao Su sharing insights on training free generation using 3d diffusion. C3DV workshop at CVPR.

Prof. Angel Chang of SFU sharing her journey on compositional scene generation at C3DV workshop #CVPR2025

Prof Tolga Birdal shared deep insights on compositionall; fun talk. C3DV workshop

Prof Yi Fang of NYU sharing latest advances geo his dean in object guided navigation at C3DV #CVPR2025

The CVPR AI Art Gallery is only with lots of amazing AI artists. Honored to have my Reflections on Mechanical Creativity included. @cvpr Luba Elliott @ CVPR

Sharing my thoughts on Environmental Conservation at The Human Capability Initiative!

We are thrilled to announce that CV for Ecology Workshop is returning for its second year at #ICCV2025 in Honolulu, Hawaii! If your work combines computer vision and ecology, submit a paper and join us! Deadlines: July 4 (Proceedings) / July 22 (Non-Archival)

🚀 Excited to share that 7 papers were accepted to ICCV 2025 main conference! 🎉 Huge congrats to the brilliant students and collaborators 🙌 — more details to follow. 📚✨!

(2/n) AVTrustBench: Assessing and Enhancing Reliability and Robustness in Audio-Visual LLMs We introduce the Audio-Visual Trustworthiness Benchmark and CAVPref, a model-agnostic training strategy that boosts AVLLM robustness and comprehension by up to 30.19% across 9 challenging

#ICML2025 XIAOQIAN SHEN and team are presenting LongVU paper at ICML on the 17th this week. The developed mechanism was shown to significantly reduce spatiotemporal redundancy in video LLMs.