🏜️Omni Driving Models🏜️

👉OmniNWM is a unified panoramic navigation world model that advances autonomous driving by jointly generating multi-modal states (RGB, semantics, depth, 3D occupancy), enabling precise action control & facilitating closed-loop evaluation through occupancy-based dense rewards. Repo under Apache 2.0💙

👉Review https://t.ly/ktXvz
👉Paper https://lnkd.in/eFKSZnrc
👉Project https://lnkd.in/eSDfccv8
👉Repo https://lnkd.in/efCSvjtp

🐠ITTO: Protocol for Dynamic Tracking🐠

👉ITTO by Caltech is a novel long-range tracking benchmark suite for evaluating and diagnosing tracking methods on complex and long-range motions. Repo under CC BY-NC 4.0💙

👉Review https://t.ly/tN84a
👉Paper https://arxiv.org/pdf/2510.19819
👉Project https://glab-caltech.github.io/ITTO/
👉Repo https://github.com/ilonadem/itto

🦗Character Mixing Generation🦗

👉MBZUAI unveils the first ever video-gen system able to preserve character ID, behavior & original style while generating plausible interactions between characters that have never coexisted - from cartoons (We Bare Bears, Tom & Jerry) to realistic humans (Mr. Bean, Young Sheldon)

👉Review https://t.ly/tN84a
👉Paper https://lnkd.in/dhKMwukv
👉Project https://lnkd.in/dBkJs48h
👉Repo https://lnkd.in/dw_uzgAk

🧷Generative Point Tracking w/ FM🧷

👉Generative Point Tracker (GenPT) is a novel generative framework for modelling multi-modal trajectories. Able to capture the multi-modality in point trajectories. Repo under MIT💙

👉Review https://t.ly/MMFrt
👉Paper https://arxiv.org/pdf/2510.20951
👉Project mtesfaldet.net/genpt_projpage/
👉Repo https://github.com/tesfaldet/genpt

🦄Unified Region-Level MLLM🦄

👉PixeRefers is an unified multimodal LLM framework that supports precise, region-specific understanding in both static images and dynamic videos, overcoming the holistic, scene-level bias of prior MLLMs. SOTA results. Demo, Repo & Dataset available💙

👉Review https://t.ly/WH4dQ
👉Paper arxiv.org/pdf/2510.23603
👉Project circleradon.github.io/PixelRefer
👉Repo https://github.com/alibaba-damo-academy/PixelRefer

🌱PlanarTrack: Large Planar Tracking🌱

👉PlanarTrack is a large-scale HQ and challenging benchmark for planar tracking: 1,150 sequences with 733K+ frames, including 1,000 short-term & 150 long-term videos. Repo & Dataset available💙

👉Review https://t.ly/mYNi7
👉Paper arxiv.org/pdf/2510.23368
👉Repo https://lnkd.in/edb3GMyT
👉Project https://lnkd.in/eC-hVB-U
👉Data https://lnkd.in/eew2j4tM

👢Generative View Stitching 👢

👉GVS is a novel approach that enables collision-free camera-guided video generation for predefined trajectories, it's a non-autoregressive alternative to video length extrapolation. Full repo under MIT💙

👉Review https://t.ly/TiN_5
👉Paper https://arxiv.org/pdf/2510.24718
👉Project https://andrewsonga.github.io/gvs/
👉Repo github.com/andrewsonga/generative_view_stitching

Greetings from the SMART CITY WORLD CONGRESS in Barcellona. If you are around, ping me ;)

🔪Tracking Object Transformations🔪

👉"Track Any State": tracking objects through transformations while detecting/describing state changes. Repo & Dataset available under MIT💙

👉Review https://t.ly/NPyW4
👉Paper https://lnkd.in/d4pA3bXJ
👉Project https://lnkd.in/dgbNfCuj
👉Repo https://lnkd.in/dtVWq2z7

🔥🔥 Sunday mood 🔥🔥

🎸Another BRIXEL in the Wall 🎸

👉BRIXEL allows the user to produce high-resolution feature maps using the DINOv3 backbone without requiring large amounts of compute. Repo released💙

👉Review https://t.ly/fZPwC
👉Paper arxiv.org/pdf/2511.05168
👉Repo github.com/alexanderlappe/BRIXEL

🐼Pixel-Dense Embedding🐼

👉FlowFeat is a novel high-resolution and multi-task feature representation that embeds a distribution of plausible apparent motions, or motion profiles. Repo available under 💙

👉Review https://t.ly/aUx_U
👉Paper arxiv.org/pdf/2511.07696
👉Project tum-vision.github.io/flowfeat
👉Repo github.com/tum-vision/flowfeat

🚨 Announcement 🚨

I’ve received numerous reports of people blatantly copying my content on LinkedIn just to get a few likes.

Let me be very clear: I put a great deal of time and effort into reviewing papers and creating original, meaningful content. It’s disappointing to see professionals (some of whom are even members of this group or my connections) resorting to plagiarism instead of contributing their own ideas.

👉 Starting today, I’ll be removing these connections from LinkedIn and banning such individuals from this group.

📢 I also encourage everyone to report these cases whenever you come across them. Every single report helps stop this bad habit and keeps our community fair, respectful, and authentic.

🟩 Foundational Humanoid 🟩

👉#NVIDIA unveils SONIC a novel foundational model for high-precision teleoperation & interactive control capabilities (running, jumping, crawling) with natural human-like movements. Code announced💙

👉Review https://t.ly/_3wnt
👉Paper https://lnkd.in/dctfShu8
👉Project https://lnkd.in/d_inmA2p

🔥Depth Anything 3 is out🔥

👉ByteDance unveils Depth Anything 3 (DA3), a model that predicts spatially consistent geometry from arbitrary visual inputs, with or without known camera poses. Repo under Apache 2.0💙

👉Review https://t.ly/AOPu7
👉Paper arxiv.org/pdf/2511.10647
👉Project https://lnkd.in/dnByyn2z
👉Repo https://lnkd.in/daCVz_4a
👉Demo https://lnkd.in/dKUZiJt

🌩️ It's "Time-to-Move" 🌩️

👉Technion + Nvidia Time-to-Move (TTM) is a training-free, plug-and-play framework for motion- and appearance-controlled video generation with I2V diffusion models (Wan 2.2, CogVideoX, & Stable VD). Impressive results!

👉Review https://t.ly/0pwXm
👉Paper https://lnkd.in/dxD3uHYb
👉Project https://lnkd.in/dcE5juyM
👉Repo https://lnkd.in/dMMUjybJ

⌚ Multi-Shot Video Segmentation ⌚

👉Fudan focuses on an underexplored task of multi-shot video object segmentation (MVOS). Benchmark and repo available (the extension part of SAM) under Apache 2.0💙

👉Review https://t.ly/WBW00
👉Paper https://arxiv.org/pdf/2511.13715
👉Project https://henghuiding.com/SAAS/
👉Repo https://github.com/FudanCVL/SAAS

🔥 SAM 3/3D are OUT!! 🔥

👉#META released SAM 3, a unified model for detection, segmentation, tracking of objects in images & video using text, exemplar & visual prompts. Repo/Models under proprietary license💙

👉Review https://t.ly/lnRZN
👉Paper https://t.ly/5tq9N
👉Project https://ai.meta.com/sam3/
👉Demo: https://segment-anything.com
👉Repo https://github.com/facebookresearch/sam3

🍯Unwrapping of 3D Meshes🍯

👉PartUV is a novel part-based UV unwrapping method for 3D meshes; it combines learned part priors with geometric cues to generate a compact set of part-aligned charts. Repo released💙

👉Review https://t.ly/8dNIY
👉Paper arxiv.org/pdf/2511.16659
👉Project www.zhaoningwang.com/PartUV/
👉Repo github.com/EricWang12/PartUV