Vincent Qin

Image matching webui is now deployed on HF, link: huggingface.co/spaces/Realcat…

Introducing LoMa, the next generation of feature matcher!

Fisheye3R: Adapting Unified 3D Feed-Forward Foundation Models to Fisheye Lenses Ruxiao Duan, Erin Hong, Dongxu Zhao, Eric Turner, Alex Wong, Yunwen Zhou tl;dr: in title arxiv.org/abs/2603.28896

TerraSky3D: Multi-View Reconstructions of European Landmarks in 4K Mattia D'Urso, Yuxi Hu, Christian Sormann, Mattia Rossi, Friedrich Fraundorfer tl;dr: new 3D dataset arxiv.org/abs/2603.28287

project page: icetea-cv.github.io/mv-roma

thin and clean point cloud😆

@gabriberton @GoogleDeepMind Congratulations, Gabriele🚀

I have joined @GoogleDeepMind! I'll be training VLMs And I'll still keep posting about latest developments on AI, Computer Vision and LLMs So no more posts on PyTorch tricks. I might post about JAX. Stay tuned...

@gabriberton 🚀🚀🚀Great!

VisMatch is on pypi! VisMatch is a wrapper for image matching models, like LightGlue, RoMa-v2, MASt3R, LoFTR, and 50+ more! It's literally as simple as: pip install vismatch vismatch-match --inputs img0 img1 --matcher choose_any To run image matching on any 2 images [1/4]

FrameVGGT: Frame Evidence Rolling Memory for streaming VGGT Zhisong Xu, Takeshi Oishi tl;dr: not token-level compression, but block-level bounded retention arxiv.org/abs/2603.07690

Code: github.com/VAISR/OVGGT

@xiaohu MiniMax M2.5呢?

OpenClaw AI Agent 小龙虾能力排行榜 专门测试各家大模型在 OpenClaw 框架下执行实际编码任务的成功率。 用一套标准化的 OpenClaw Agent 任务来跑各个模型，通过自动化检查 + LLM 评审来打分，衡量每个模型完成任务的成功率。 前三名分别为： Gemini 3 Flash Preview MiniMax M2.1 Kimi K2.5 然后是： Claude Sonnet 4.5 Gemini 3 Pro Preview Claude Haiku 4.5 Claude Opus 4.6 Claude 家族三个模型都在 90% 以上，GPT-5.2 反而只有 65.6% 排名靠后，DeepSeek V3.2 在 82% 左右。

What happens when VLMs meet 3D foundation models? See VLM-3R (CVPR 2026). VLM-3R links a vision-language model (e.g., Qwen) with 3D geometric foundation models (e.g., CUT3R) at metric scale. Given an uncalibrated video, it moves beyond pixels to perceive and reason in 3D space. Code (open source): vlm-3r.github.io

ZipMap: Linear-Time Stateful 3D Reconstruction with Test-Time Training @Haian_Jin, Rundi Wu, @tianyuanzhang99, @RuiqiGao, @jon_barron, Noah Snavely, @holynski_ tl;dr: another(?) TTT+VGGT arxiv.org/abs/2603.04385

Code: github.com/weitong8591/gl…

Code: github.com/ywh187/XStream…

VLG-Loc Vision-Language Global Localization (VLG-Loc) is a global localization method that uses camera images and a human-readable labeled footprint map containing only names and areas of distinctive visual landmarks github.com/CyberAgentAILa…

@yzly1 @zhenjun_zhao The author has withdrawn the paper.

@zhenjun_zhao I think the results of XFeat and RIPE on MegaDepth seem unusually low. Are these the originally reported results, or were they obtained under different settings?

From Pairs to Sequences: Track-Aware Policy Gradients for Keypoint Detection Yepeng Liu, Hao Li, Liwen Yang, Fangzhen Li, Xudi Ge, Yuliang Gu, kuang Gao, Bing Wang, Guang Chen, Hangjun Ye, Yongchao Xu tl;dr: multi-view version of RL-based method (RFP/RIPE) for detection; RDD as backbone no eval. on IMC arxiv.org/abs/2602.20630

OpenVO: Open-World Visual Odometry with Temporal Dynamics Awareness @phucnda, @anh_n_nhu, @MingCLinCS tl;dr: temporal dynamics+scene geometry arxiv.org/abs/2602.19035

Have We Mastered Scale in Deep Monocular Visual SLAM? The ScaleMaster Dataset and Benchmark Hyoseok Ju, Bokeon Suh, @GiseopK tl;dr: in title arxiv.org/abs/2602.18174

🚀 #CVPR2026 Accepted!🚀 Thrilled to share that my first-authored undergraduate paper, “Emergent Extreme-View Geometry in 3D Foundation Models,” has been accepted to CVPR 2026! 🎉 Looking forward to seeing many of you in Denver! ✈️ Project page: ext-3dfms.github.io