🐞6D Object Pose w/ Deformation🐞

👉DeSOPE by Xidian & #MagicLeap is a novel large-scale dataset for 6DoF deformed objects: 665K pose annotations produced via a semiautomatic pipeline. Repo & Dataset announced💙

👉Review https://t.ly/M5VgX
👉Paper https://arxiv.org/pdf/2604.06720
👉Project https://desope-6d.github.io/
👉Repo TBA

🔥SOTA 3D Detection in the wild🔥

👉WildDet3D is a novel unified geometry-aware architecture for 3D detection that natively accepts text, point, and box prompts and can incorporate auxiliary depth signals at inference time. New SOTA! Repo, models and iphone 💙

👉Review https://t.ly/8NxBN
👉Paper arxiv.org/pdf/2604.08626
👉Project allenai.github.io/WildDet3D/
👉Repo github.com/allenai/WildDet3D

🧴OmniShow Content Creation🧴

👉OmniShow is the novel SOTA in content creation with industry-grade performance. Impressive results, best with audio. Repo announced💙

👉Review https://t.ly/Pm-7U
👉Paper arxiv.org/pdf/2604.11804
👉Project correr-zhou.github.io/OmniShow/
👉Repo github.com/Correr-Zhou/OmniShow

🐓Interactive Objects from EgoVideo🐓

👉EgoFun3D by Simon Fraser University is a coordinated task, dataset and benchmark for modeling interactive 3D objects from egocentric videos. Repo (TBA), demo & dataset💙

👉Review https://t.ly/YhGN7
👉Paper arxiv.org/pdf/2604.11038
👉Project 3dlg-hcvc.github.io/EgoFun3D/
👉Repo github.com/3dlg-hcvc/EgoFun3D
👉Demo bc79fea884062374b3.gradio.live/

📱3D Human-Object Contact📱

👉Pi-HOC by CMU + NREC is a novel single-pass, instance-aware framework for dense 3D semantic contact prediction of all human-object pairs. Repo announced💙

👉Review https://t.ly/TAgG1
👉Paper https://arxiv.org/pdf/2604.12923
👉Project https://pi-hoc.github.io/
👉Repo https://github.com/SravanChittupalli/Pi-HOC

🐞GCT 3D Reconstruction🐞

👉ANT unveils LingBot-Map, a feed-forward 3D foundation model for reconstructing scenes from streaming data, built upon a geometric context transformer (GCT) architecture. Repo under A-NC 4.0 International💙

👉Review https://t.ly/ExodA
👉Paper https://arxiv.org/pdf/2604.14141
👉Project https://arxiv.org/pdf/2604.14141
👉Repo github.com/robbyant/lingbot-map

👩‍🦰Deformable 3D Hair👩‍🦰

👉Xi’an Jiaotong University unveils a novel method that reconstructs decoupled 3D Gaussian head avatars from a single input image: effortless hairstyle transfer with natural dynamic hair motion. Code announced💙

👉Review https://t.ly/kWZdd
👉Paper https://arxiv.org/pdf/2604.14782
👉Project yuansun-xjtu.github.io/CompHairHead.io/
👉Repo yuansun-xjtu.github.io/CompHairHead.io/

🌗Mobile Ultra-detailed Avatars🌗

👉Given skeletal poses and a virtual camera as inputs, MUA by Max Planck Institute produces photorealistic renderings and hyper-detailed geometry of animatable clothed humans. Repo announced💙

👉Review https://t.ly/QPCy6
👉Paper https://arxiv.org/pdf/2604.18583
👉Project https://vcai.mpi-inf.mpg.de/projects/MUA/
👉Repo TBA

🎈Face Anything 4D (SOTA)🎈

👉A novel unified 4D facial reconstruction and dense tracking from image sequences: new SOTA in facial single-image and mono-video depth estimation, dense 4D reconstruction, and 3D point tracking. Repo & Dataset announced💙

👉Review https://t.ly/zItie
👉Paper https://arxiv.org/pdf/2604.19702
👉Project kocasariumut.github.io/FaceAnything
👉Repo TBA

💙 PY4AI 2026: here we are! 💙

👉The third edition of our conference is official! Speaker list and (free) tickets: https://t.ly/L4_52

🛒 Reshoot-Anything is out 🛒

👉Reshoot-Anything reshoots dynamic monocular videos under novel camera trajectories. Code under Apache 2.0 💙

👉Review https://t.ly/MIqAc
👉Paper https://arxiv.org/pdf/2604.21776
👉Project adithyaiyer1999.github.io/reshoot-anything/
👉Repo github.com/morphicfilms/video-to-video

🧘‍♀️Holistic Shot Boundary Detection🧘‍♀️

👉OmniShotCut detects shot changes of the video in diverse sources (anime, vlog, game, shorts, sports, screen recording, etc.), and recognize Sudden Jump and Transitions (dissolve, fade, wipe, etc.) by proposing a Shot-Query-based Video Transformer. Repo, demo & benchmark💙

👉Review https://t.ly/sTi7N
👉Paper https://arxiv.org/pdf/2604.24762
👉Project uva-computer-vision-lab.github.io/OmniShotCut_website/
👉Repo github.com/UVA-Computer-Vision-Lab/OmniShotCut

🪝Syn4D: Multiview Synthetic 4D Dataset🪝

👉Syn4D is novel multi-view synthetic dataset of dynamic scenes that includes ground-truth camera motion, depth maps, dense tracking, and parametric human pose annotations💙

👉Review https://t.ly/SL1mk
👉Paper https://arxiv.org/pdf/2605.05207
👉Project https://jzr99.github.io/Syn4D/
👉Repo https://github.com/jzr99/Syn4D
👉Data huggingface.co/datasets/Syn4D/Syn4D_RGBD/tree/main

🦄Unified Correspondence Transformer🦄

👉UniCorrn is the first correspondence model with shared weights that unifies 2D-2D, 2D-3D, and 3D-3D geometric matching with a transformer. CC BY-NC-SA 4.0💙

👉Review https://t.ly/2OBdq
👉Paper https://arxiv.org/pdf/2605.04044
👉Project https://neu-vi.github.io/UniCorrn/
👉Repo https://github.com/neu-vi/UniCorrn

🍒Count Anything, Any Granularity🍒

👉Open-world counting as multi-grained counting, where visual exemplars specify target appearance and fine-grained text specifies the intended semantic granularity across five explicit levels. Repo/Data under Apache💙

👉Review https://t.ly/nqz80
👉Paper https://lnkd.in/dp7khTRU
👉Project https://lnkd.in/d_jfX_Yn
👉Repo https://lnkd.in/dkTRGZkG
👉Data https://lnkd.in/dB83jRyT

🪔Latent Decoding Pixel Diffusion🪔

👉PiD by Nvidia is a plug-and-play diffusion decoder that replaces VAE/RAE decoders, turning latent representations directly into super-resolved pixels in a single pass. Repo under Apache 2.0💙

👉Review https://t.ly/y19mA
👉Paper https://lnkd.in/duVC25C2
👉Project https://lnkd.in/dW6TkzCB
👉Repo https://lnkd.in/dnGdgKRr

🔍 Nvidia Locate Anything 🔍

👉Diverse localization tasks under a unified vision-language model, including document understanding, GUI grounding, dense detection, and OCR. Repo released💙

👉Review https://t.ly/PvwFo
👉Paper https://lnkd.in/dWfNpzPZ
👉Project https://lnkd.in/dM89BX-8
👉Repo https://lnkd.in/dC4KCQSM

🕷️Human Universal Grasping🕷️

👉HUG is a flow-matching model that generates diverse human grasps for any user-specified object in a single RGB-D image captured from a stereo camera.

👉Review https://t.ly/VG1Eu
👉Paper https://arxiv.org/pdf/2606.17054
👉Repo https://github.com/KevinyWu/hug
👉Project https://grasping.io/

🔊VolHuMe - Volumetric Human Meshes🔊

👉VolHuMe (H/T @Martinella_94) is a novel, high-resolution large-scale dataset of volumetric human meshes with complete 4D GT: multi-view RGB-D, textured meshes, dense point clouds, normal maps, rigged assets, garment segmentation, and SMPL-X fittings in one dataset. Insane💙

👉Review https://t.ly/b5vxy
👉Paper https://arxiv.org/pdf/2606.23062
👉Project giuli13.github.io/volhume-website/#
👉Repo TBA soon