北京时间7月8日晚上8点，南洋理工大学MMLab博士生吴鹏浩，将直播分享《GUI-Reflection：让多模态 GUI 智能体获得反思纠错能力的训练框架》。

👁️‍ ObjectClear: Complete Object Removal via Object-Effect Attention 🧹 Jupyter Notebook 🥳 Thanks to Jixin Zhao ❤ Shangchen Zhou ❤ Zhouxia Wang ❤ Peiqing Yang ❤ Chen Change Loy ❤ 🌐page: zjx0101.github.io/projects/Objec… 🧬code: github.com/zjx0101/Object… 📄paper: arxiv.org/abs/2505.22636

🔥🆕 ObjectClear is an object removal model that can jointly eliminate the target object and its associated effects (shadow etc) Object Clear app on @Huggingface : huggingface.co/spaces/jixin01…

Introducing 𝗦𝗧𝗿𝗲𝗮𝗺𝟯𝗥, a new 3D geometric foundation model for efficient 3D reconstruction from streaming input. Similar to LLMs, STream3R uses casual attention during training and KVCache at inference. No need to worry about post-alignment or reconstructing from scratch.

STream3R Scalable Sequential 3D Reconstruction with Causal Transformer

Our new preprint: “Next Visual Granularity Generation”  -  a novel framework for image generation that builds visuals hierarchically, from broad layout to fine detail. Achieves consistent FID improvements (e.g., from 3.30 → 3.03, 2.57 → 2.44, 2.09 → 2.06) compared to VAR in

STream3R reformulates dense 3D reconstruction into a sequential registration task with causal attention. Just tried 3D reconstruction on a #GrokImagine video using #STream3R🫡! Check out STream3R on our GitHub for more👨‍💻: github.com/NIRVANALAN/STr…

Next Visual Granularity Generation

New from S-Lab, Nanyang Technological University & SenseTime Research: Next Visual Granularity Generation (NVG)! This novel framework progressively refines images from global layout to fine details, offering fine-grained control over generation. It outperforms the VAR series in

Lan and Luo et al., "STREAM3R: Scalable Sequential 3D Reconstruction with Causal Transformer" Yep, another streaming feed-forward 3D estimator. This time, with Dust3R backbone. Architecture is now getting pretty close to LLMs :) Are these going to become 3D GPT?

Congrats to Ziwei Liu !!

📸Join us at #ICCV2025 for the Mobile Intelligent Photography & Imaging (MIPI) Workshop! ✨Leading keynotes: Profs. Song Han, Michal Irani, Boxin Shi, and Ming-Hsuan Yang - on intelligent photography and efficient GenAI. 🗓Oct 20, 8:50am–12:30pm HST 🔗mipi-challenge.org

Thinking with Camera A Unified Multimodal Model for Camera-Centric Understanding and Generation

Introducing 𝐓𝐡𝐢𝐧𝐤𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐂𝐚𝐦𝐞𝐫𝐚📸, a unified multimodal model that integrates camera-centric spatial intelligence to interpret and create scenes from arbitrary viewpoints. Project Page: kangliao929.github.io/projects/puffi… Code: github.com/KangLiao929/Pu…

Thinking with Camera (Puffin) just dropped. This AI doesn't just see a picture, it reasons like a director. It predicts lens/pose, guides shots, and generates scenes across views. Simple breakdown:

The #CVPR2026 submission portal is now open!

Congrats to Yuekun Dai Yuekun Dai and Ziang Cao ziangc , both from MMLab@NTU , for winning the prestigious Google PhD Fellowship! Yuekun: ykdai.github.io Ziang: ziangcao0312.github.io NTU Singapore

🔥One-Stop Training Engine for Unified Models🔥 ⚡️LMMs-Engine⚡️ is a lean and flexible unified model training engine built for hacking at scale * Support multimodal inputs and outputs, from AR, diffusion and linear models, to unified models like BAGEL 🏠github.com/EvolvingLMMs-L…