🍡 DWPose: 2-stage Pose Distillation 🍡

👉 Tsinghua (+IDEA) unveils a novel two-stage pose Distillation for whole-body pose estimation.

😎Review https://t.ly/BSi20
😎Paper arxiv.org/pdf/2307.15880.pdf
😎Code github.com/IDEA-Research/DWPose

👗 Multimodal Neural Designer 👗

👉 Multimodal #AI that can generate novel fashion images conditioned on text, keypoints, and sketches

😎Review https://t.ly/zVk70
😎Paper arxiv.org/pdf/2304.02051.pdf
😎Code github.com/aimagelab/multimodal-garment-designer

📸 Computational Burst Photography in App 📸

👉#Google unveils a novel computational burst system to democratize the professional photography via smartphone

😎Review https://t.ly/5ibJX
😎Paper arxiv.org/pdf/2308.01379.pdf
😎Project https://motion-mode.github.io

🎠Neural Closed-Loop Simulator🎠

👉A neural sensor simulator that takes a single recorded log captured by a sensor-equipped vehicle and converts it into a realistic closed-loop multi-sensor simulation

😎Review https://t.ly/EcRLc
😎Paper arxiv.org/pdf/2308.01898.pdf
😎Project https://waabi.ai/unisim/

🙏 A quick poll for helping me in improving the quality of the contents about #computervision.

Please give me a feedback here: https://t.ly/qXb4C

Thanks :)

🪛 HANDAL: Real-World Manipulable Objects 🪛

👉 #Nvidia unveils HANDAL dataset: category-level object pose and affordance prediction

😎Review https://t.ly/MXZDI
😎Paper arxiv.org/pdf/2308.01477.pdf
😎Dataset wenbowen123.github.io/handaldataset

🎨 Interactive Neural Painting 🎨

👉 Novel AI-powered tool to help artists in completing their artworks

😎Review https://t.ly/ELUb0
😎Paper arxiv.org/pdf/2307.16441.pdf
😎Project helia95.github.io/inp-website
😎Supp helia95.github.io/inp-website/supp_mat.html

👩‍🚀 HD Avatar via Text & Pose 👩‍🚀

👉 Generating expressive #3D avatars from nothing but text descriptions & pose guidance

😎Review https://t.ly/wrSMH
😎Paper arxiv.org/pdf/2308.03610.pdf
😎Project avatarverse3d.github.io

🐘 Controllable Synthetic Data (extending Image-Net) 🐘

👉#META's PUG, a new generation of interactive environments for representation learning. Extending Image-Net!

😎Review https://t.ly/nCYs0
😎Paper arxiv.org/pdf/2308.03977.pdf
😎Project pug.metademolab.com
😎Code github.com/facebookresearch/PUG

🔥The code is out!🔥

😎 Code https://github.com/NVlabs/neuralangelo

🌈 Tracking by Persistent Dynamic View Synthesis 🌈

👉Novel simultaneous addressing of dynamic scene novel-view synthesis + 6-DOF tracking of all dense scene elements

😎Review https://t.ly/Bc535
😎Paper arxiv.org/pdf/2308.09713.pdf
😎Project dynamic3dgaussians.github.io
😎Code github.com/JonathonLuiten/Dynamic3DGaussians

🛒 Digital Twins for AutoRetail Checkout 🛒

👉From #Nvidia a novel approach for using 3D assets for training 2D detection and tracking model in AutoRetail Checkout

😎Review https://t.ly/Ea7kt
😎Paper arxiv.org/pdf/2308.09708.pdf
😎Code github.com/yorkeyao/Automated-Retail-Checkout

🥎SportsMOT + MixSort = Sport MOT🥎

👉Nanjing just released a MOT dataset for sports scenes + the SOTA code/model for tracking (MixSort)

😎Review https://t.ly/NHUxL
😎Paper arxiv.org/pdf/2304.05170.pdf
😎Code github.com/MCG-NJU/MixSort
😎Project deeperaction.github.io/datasets/sportsmot.html

⚡️Feature Matching at Light Speed⚡️

👉LightGlue is a lightweight feature matcher with high accuracy and blazing fast inference

😎Review https://t.ly/jkecX
😎Paper arxiv.org/pdf/2306.13643.pdf
😎Code github.com/cvg/LightGlue

🕹️ CoDeF: Video Content Deformation Fields 🕹️

👉CoDeF is a new type of video representation for video-editing tasks

😎Review https://t.ly/PIVl-
😎Paper arxiv.org/pdf/2308.07926.pdf
😎Project https://qiuyu96.github.io/CoDeF
😎Code https://github.com/qiuyu96/CoDeF

Hello everybody,
a lot of you asked me to open the comments to better enjoy the posts. I want to follow your suggestion, hope you will enjoy this new mood!

🔥 NO SPAM
🔥 NO COMMERCIAL
🔥 NO UNRESPECTFUL MESSAGEs

🧡JUST AI & SCIENCE

⚠️ BAN AT THE FIRST VIOLATION ⚠️

🦠 Instance-Level Semantics of Cells 🦠

👉TYC: novel dataset for understanding instance-level semantics & motions of cells in microstructures

😎Review https://t.ly/y-4VZ
😎Paper arxiv.org/pdf/2308.12116.pdf
😎Project christophreich1996.github.io/tyc_dataset/
😎Code github.com/ChristophReich1996/TYC-Dataset
😎Data tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/3930

🌵POCO: 3D HPS + Confidence🌵

👉 Novel framework for HPS: #3D human body + confidence in a single feed-forward pass

😎Review https://t.ly/cDePe
😎Paper arxiv.org/pdf/2308.12965.pdf
😎Project https://poco.is.tue.mpg.de