🫀CPR-Coach: Neural Cardiopulmonary Resuscitation🫀

👉CPR-Coach: fine-grained action recognition in cardiopulmonary resuscitation

😎Review https://t.ly/Qbg4K
😎Paper arxiv.org/pdf/2309.11718.pdf
😎Code github.com/Shunli-Wang/CPR-Coach
😎Project shunli-wang.github.io/CPR-Coach

🧪 NeuralLabeling with NeRF 🧪

👉Annotating a scene by generating segmentation masks, affordance maps, 2D bounding boxes, 3D BB, 6DOF poses, depth & meshes.

😎Review https://t.ly/1GPsj
😎Paper arxiv.org/pdf/2309.11966.pdf
😎Code github.com/FlorisE/neural-labeling
😎Project florise.github.io/neural_labeling_web

🍟 DE-ViT: detecting everything via DINOv2 🍟

👉DE-ViT: open-set object detector based on DINOv2 backbone. It's the new SOTA on COCO & LVIS dataset

😎Review https://t.ly/_DAmt
😎Paper arxiv.org/pdf/2309.12969.pdf
😎Code https://github.com/mlzxy/devit

🛵CoTracker: fast transformer-tracker🛵

👉META's CoTracker is a fast transformer-based model that can track any point in a video

😎Review https://t.ly/M36A_
😎Paper arxiv.org/pdf/2307.07635.pdf
😎Project https://co-tracker.github.io/
😎Code github.com/facebookresearch/co-tracker

🌬️ Neural Blowing in Still Photos 🌬️

👉 A novel approach to animate human hair (and clothes) in a still portraits

😎Review https://t.ly/HKG0t
😎Paper arxiv.org/pdf/2309.14207.pdf
😎Project nevergiveu.github.io/AutomaticHairBlowing

🌮 OW Indoor Segmentation 🌮

👉3D-OWIS is a novel open-world 3D indoor instance segmentation method (with auto-labeling scheme) to separate known/unknown category labels

😎Review https://t.ly/-7ALf
😎Paper arxiv.org/pdf/2309.14338.pdf
😎Code github.com/aminebdj/3D-OWIS

🧱 Generating Scenes from Touch 🧱

👉#AI for synthesizing images from tactile signals (and vice versa) and apply it to a number of visuo-tactile synthesis tasks

😎Review https://t.ly/Gxr0L
😎Paper https://arxiv.org/pdf/2309.15117.pdf
😎Project https://fredfyyang.github.io/vision-from-touch
😎Code https://github.com/fredfyyang/vision-from-touch

☕Decaf: 3D Face-Hand Interactions☕

👉The first learning-based MoCap to track human hands interacting with human faces in #3D from single monocular RGB videos

😎Review https://t.ly/070Tj
😎Paper arxiv.org/pdf/2309.16670.pdf
😎Project vcai.mpi-inf.mpg.de/projects/Decaf

🌱 Making LLaMA See and Draw 🌱

👉Tencent #AI planted a SEED of Vision in Large Language Model. Making LLaMA see 'n' draw stuff.

😎Review https://t.ly/QiCAv
😎Paper arxiv.org/pdf/2310.01218.pdf
😎Code github.com/AILab-CVC/SEED

🔥Visual-Math Q&A: MathVista is out! 🔥

👉 MathVista is the ultimate benchmark designed to amalgamate challenges from diverse mathematical and visual tasks

😎Review https://t.ly/yfqHZ
😎Paper https://arxiv.org/pdf/2310.02255.pdf
😎Project https://mathvista.github.io/
😎Code github.com/lupantech/MathVista

💚💙 Where Is OpenCV 5? 💙💚

👉On October 24th, the organization is launching a crowdfunding campaign to raise funds for #OpenCV 5 development.

👆me in 2008 during my thesis work about face tracking; up to 50x faster than the previous SOTA. No chance to did it without OpenCV library and support from the community.

🔥Support #OpenCV 5 to create the next-gen of researchers and scientists. Spread the voice: https://t.ly/UTukV

🏊SwimXYZ: Synthetic Swim🏊

👉SwimXYZ: synthetic dataset for swimming, monocular videos annotated with ground truth 2D and 3D joints

😎Review https://t.ly/F-rdF
😎Paper arxiv.org/pdf/2310.04360.pdf
😎Data g-fiche.github.io/research-pages/swimxyz

📊 TextPSG: PSG from Text 📊

👉A novel problem in #AI: Panoptic Scene Graph Generation from Purely Textual Descriptions (Caption-toPSG)

😎Review https://t.ly/UXEmk
😎Paper arxiv.org/pdf/2310.07056.pdf
😎Project vis-www.cs.umass.edu/TextPSG
😎Code github.com/chengyzhao/TextPSG

🙋 Full Human Motion 🙋

👉OmniControl by Google is novel framework for text-conditioned human motion generation model based on diffusion process

😎Review https://t.ly/F_0Ov
😎Paper arxiv.org/pdf/2310.08580.pdf
😎Project neu-vi.github.io/omnicontrol/

🦹‍♀️ Snap's Hyper-Realistic Human 🦹‍♀️

👉New diffusive #AI by Snap that generates in-the-wild human images with hyper-realism. Swipe the gallery, NUTS!👇

😎Gallery https://t.ly/cG74X
😎Paper arxiv.org/pdf/2310.08579.pdf
😎Project snap-research.github.io/HyperHuman
😎Code github.com/snap-research/HyperHuman

👗AG3D clothed avatar from 2D👗

👉The novel SOTA in adversarial generative of realistic 3D people

😎Review https://t.ly/vnJO7
😎Project https://zj-dong.github.io/AG3D
😎Code https://github.com/zj-dong/AG3D
😎Paper zj-dong.github.io/AG3D/assets/paper.pdf

🌱Pose-Format: All-in-One Pose🌱

👉 Pose-format: a comprehensive toolkit designed for human pose: unified, flexible, and easy-to-use

😎Review https://t.ly/rFrhq
😎Paper arxiv.org/pdf/2310.09066.pdf
😎Code github.com/sign-language-processing/pose

😻 CatFLW: Cat Neural Landmarks 😻

👉Landmark convolution neural network-based model for cat faces

😎Review https://t.ly/Y3mQ8
😎Paper arxiv.org/pdf/2305.04232.pdf
😎Dataset www.tech4animals.org/catflw

🍡4K4D: Real-Time 4D at 4K🍡

👉THE new SOTA in view synthesis of dynamic 3D scenes at 4K. 30x faster, up to 400 FPS. Nuts!

😎Review https://t.ly/6ddQh
😎Paper arxiv.org/pdf/2310.11448.pdf
😎Project zju3dv.github.io/4k4d/
😎Code github.com/zju3dv/4K4D

🛣️ Holistic Parking Detection (YOLO) 🛣️

👉 One-step Holistic Parking Slot Network: a tailor-made adaptation of YOLOv4 algorithm for all-shaped parking slot detection

😎Review https://t.ly/2l4ZG
😎Paper arxiv.org/pdf/2310.11629.pdf