This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅฌ "Perception Test" by #DeepMind ๐ฅฌ
๐Huge dataset with obj & point tracks, temporal sounds, multiple & grounded vQA
๐Review https://bit.ly/3Vqh96Q
๐Dataset github.com/deepmind/perception_test
๐Project www.deepmind.com/blog/measuring-perception-in-ai-models
๐Huge dataset with obj & point tracks, temporal sounds, multiple & grounded vQA
๐Review https://bit.ly/3Vqh96Q
๐Dataset github.com/deepmind/perception_test
๐Project www.deepmind.com/blog/measuring-perception-in-ai-models
๐15๐ฅ4๐ฑ3
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅ Matterport 3D Semantics Dataset ๐ฅ
๐#Meta opens HM3DSEM, the largest #3D real-world dataset with dense semantic
๐Review https://bit.ly/3yF4W4G
๐Paper arxiv.org/pdf/2210.05633.pdf
๐Project aihabitat.org/datasets/hm3d-semantics
๐Data github.com/matterport/habitat-matterport-3dresearch
๐#Meta opens HM3DSEM, the largest #3D real-world dataset with dense semantic
๐Review https://bit.ly/3yF4W4G
๐Paper arxiv.org/pdf/2210.05633.pdf
๐Project aihabitat.org/datasets/hm3d-semantics
๐Data github.com/matterport/habitat-matterport-3dresearch
๐13
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฆ Instant Map-free Relocalization ๐ฆ
๐#Niantic unveils a novel instant, metric scaled re-localization with one single photo
๐Review https://bit.ly/3S1Gdyh
๐Paper arxiv.org/pdf/2210.05494.pdf
๐Project research.nianticlabs.com/mapfree-reloc-benchmark
๐Data research.nianticlabs.com/mapfree-reloc-benchmark/dataset
๐#Niantic unveils a novel instant, metric scaled re-localization with one single photo
๐Review https://bit.ly/3S1Gdyh
๐Paper arxiv.org/pdf/2210.05494.pdf
๐Project research.nianticlabs.com/mapfree-reloc-benchmark
๐Data research.nianticlabs.com/mapfree-reloc-benchmark/dataset
๐ฅ13๐2
This media is not supported in your browser
VIEW IN TELEGRAM
๐งฎ Novel DM for 3D Shapes by #Nvidia ๐งฎ
๐Hierarchical Latent Point Diffusion Model (LION) for 3D shape generation
๐Review https://bit.ly/3yDhZ6I
๐Paper arxiv.org/pdf/2210.06978.pdf
๐Project https://nv-tlabs.github.io/LION/
๐Code(soon) github.com/nv-tlabs/LION
๐Hierarchical Latent Point Diffusion Model (LION) for 3D shape generation
๐Review https://bit.ly/3yDhZ6I
๐Paper arxiv.org/pdf/2210.06978.pdf
๐Project https://nv-tlabs.github.io/LION/
๐Code(soon) github.com/nv-tlabs/LION
โค11๐ฑ2๐ฅ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ชฒ#6D estimation fully in the wild๐ชฒ
๐First ever self-supervised 6D pose estimation training in the wild
๐Review https://bit.ly/3yHdHuS
๐Paper arxiv.org/pdf/2210.07199.pdf
๐Project kywind.github.io/self-pose
๐Code (soon)
๐First ever self-supervised 6D pose estimation training in the wild
๐Review https://bit.ly/3yHdHuS
๐Paper arxiv.org/pdf/2210.07199.pdf
๐Project kywind.github.io/self-pose
๐Code (soon)
๐15๐คฏ8๐ฑ4
This media is not supported in your browser
VIEW IN TELEGRAM
โฝ Stable Diffusion in #Blender โฝ
๐Render with SuperPowers: novel scene render via text prompt
๐Review https://bit.ly/3s1mEeN
๐Code github.com/benrugg/AI-Render
๐Render with SuperPowers: novel scene render via text prompt
๐Review https://bit.ly/3s1mEeN
๐Code github.com/benrugg/AI-Render
๐คฏ8๐5โค2
This media is not supported in your browser
VIEW IN TELEGRAM
โฝMarkerless Body-Object Interactionโฝ
๐Novel whole-bodies/objects interaction method from multi-view RGB-D data
๐Review https://bit.ly/3yO56GY
๐Data intercap.is.tue.mpg.de/login.php
๐Project https://intercap.is.tue.mpg.de
๐Code github.com/YinghaoHuang91
๐Paper intercap.is.tue.mpg.de/media/upload/main.pdf
๐Novel whole-bodies/objects interaction method from multi-view RGB-D data
๐Review https://bit.ly/3yO56GY
๐Data intercap.is.tue.mpg.de/login.php
๐Project https://intercap.is.tue.mpg.de
๐Code github.com/YinghaoHuang91
๐Paper intercap.is.tue.mpg.de/media/upload/main.pdf
๐ฅ6๐2๐พ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅ Dressing Avatars by #META ๐ฅ
๐Novel deep photorealistic appearance method for physically-simulated clothing in #metaverse
๐Review https://bit.ly/3yRBW9Y
๐Paper arxiv.org/pdf/2206.15470.pdf
๐Novel deep photorealistic appearance method for physically-simulated clothing in #metaverse
๐Review https://bit.ly/3yRBW9Y
๐Paper arxiv.org/pdf/2206.15470.pdf
๐คฏ7๐5๐พ2โค1
AI with Papers - Artificial Intelligence & Deep Learning
๐ท๐ทOMNI3D: #3D Objects in the Wild๐ท๐ท ๐#3D detection: 234k images, 3M+ instances & 97 categories ๐๐ข๐ ๐ก๐ฅ๐ข๐ ๐ก๐ญ๐ฌ: โ
OMNI3D from publicly released dataset โ
234k pics, 3M+ annotation with 3D box โ
97 categories such as sofa, table, cars โ
Fast (450x) and exact algorithmโฆ
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅMeta Omni3D: code is out!๐ฅ
๐Source Code, models, and data just released by #META !
๐Review https://bit.ly/3MIWxD9
๐Paper arxiv.org/pdf/2207.10660.pdf
๐Project garrickbrazil.com/omni3d/
๐Code github.com/facebookresearch/omni3d
๐Source Code, models, and data just released by #META !
๐Review https://bit.ly/3MIWxD9
๐Paper arxiv.org/pdf/2207.10660.pdf
๐Project garrickbrazil.com/omni3d/
๐Code github.com/facebookresearch/omni3d
๐ฅ14๐2๐1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ช Parallel NeRF for 6-DoF pose ๐ช
๐#Nvidia unveils a parallel NeRF for 6-DoF target pose estimation
๐Review https://bit.ly/3guWWwA
๐Paper arxiv.org/pdf/2210.10108.pdf
๐Project https://pnerfp.github.io/
๐#Nvidia unveils a parallel NeRF for 6-DoF target pose estimation
๐Review https://bit.ly/3guWWwA
๐Paper arxiv.org/pdf/2210.10108.pdf
๐Project https://pnerfp.github.io/
๐8๐ฅ3
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฆLaMAR: Localization/Mapping for #AR๐ฆ
๐A new benchmark for #AR in large and unconstrained scenes
๐Review https://bit.ly/3DjlnWU
๐Paper lamar.ethz.ch/files/LaMAR.pdf
๐Project https://lamar.ethz.ch/
๐Code github.com/microsoft/lamar-benchmark
๐A new benchmark for #AR in large and unconstrained scenes
๐Review https://bit.ly/3DjlnWU
๐Paper lamar.ethz.ch/files/LaMAR.pdf
๐Project https://lamar.ethz.ch/
๐Code github.com/microsoft/lamar-benchmark
๐7๐ฅ4๐ฏ4
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅNew SOTA in Panoptic Segmentation๐ฅ
๐#Google (with Hinton๐คฏ) unveils Pix2Seq-D: novel generalist framework for panoptic segmentation
๐Review https://bit.ly/3DmpbGM
๐Paper arxiv.org/pdf/2210.06366.pdf
๐#Google (with Hinton๐คฏ) unveils Pix2Seq-D: novel generalist framework for panoptic segmentation
๐Review https://bit.ly/3DmpbGM
๐Paper arxiv.org/pdf/2210.06366.pdf
๐ฅ9๐5๐คฏ3
This media is not supported in your browser
VIEW IN TELEGRAM
๐จ UniColor: Unified Colorization ๐จ
๐The first unified framework for colorization via stroke, exemplar, text, and a mix of them
๐Review https://bit.ly/3gESR9y
๐Paper arxiv.org/pdf/2209.11223.pdf
๐Project luckyhzt.github.io/unicolor
๐Code (SOON)
๐The first unified framework for colorization via stroke, exemplar, text, and a mix of them
๐Review https://bit.ly/3gESR9y
๐Paper arxiv.org/pdf/2209.11223.pdf
๐Project luckyhzt.github.io/unicolor
๐Code (SOON)
๐คฏ18๐ฅ6๐1
This media is not supported in your browser
VIEW IN TELEGRAM
๐คฏ Full-Body from head/hand signals ๐คฏ
๐#Meta unveils AvatarPoser: first full-body pose method via userโs head/hands
๐Review https://bit.ly/3gESR9y
๐Paper arxiv.org/pdf/2207.13784.pdf
๐Code github.com/eth-siplab/AvatarPoser
๐#Meta unveils AvatarPoser: first full-body pose method via userโs head/hands
๐Review https://bit.ly/3gESR9y
๐Paper arxiv.org/pdf/2207.13784.pdf
๐Code github.com/eth-siplab/AvatarPoser
๐9๐3โค1
This media is not supported in your browser
VIEW IN TELEGRAM
๐คJRBD: Egocentric Perception of Humans๐ค
๐Stanford -> JRDB-Pose: Dataset with 600,000+ body pose annotations!
๐Review https://bit.ly/3gEZBE4
๐Paper arxiv.org/pdf/1910.11792.pdf
๐Project jrdb.erc.monash.edu/
๐Stanford -> JRDB-Pose: Dataset with 600,000+ body pose annotations!
๐Review https://bit.ly/3gEZBE4
๐Paper arxiv.org/pdf/1910.11792.pdf
๐Project jrdb.erc.monash.edu/
๐8๐ฏ4
This media is not supported in your browser
VIEW IN TELEGRAM
โ๏ธSOTA Action Detector @90+ FPS!โ๏ธ
๐YOWO-plus: real-time method for spatio-temporal action detection. YOWO-Nano the fastest!
๐Review https://bit.ly/3TUdhcI
๐Paper arxiv.org/pdf/2210.11219.pdf
๐Code github.com/yjh0410/PyTorch_YOWO
๐YOWO-plus: real-time method for spatio-temporal action detection. YOWO-Nano the fastest!
๐Review https://bit.ly/3TUdhcI
๐Paper arxiv.org/pdf/2210.11219.pdf
๐Code github.com/yjh0410/PyTorch_YOWO
๐13๐ฅฐ2
This media is not supported in your browser
VIEW IN TELEGRAM
๐๏ธ๐๏ธAutoregressive NeRF-Avatar๐๏ธ๐๏ธ
๐AutoAvatar by #Meta: autoregressive method for modeling dynamically deforming human bodies from raw scans
๐Review https://bit.ly/3W0oTgo
๐Paper arxiv.org/pdf/2203.13817.pdf
๐Project zqbai-jeremy.github.io/autoavatar
๐Code github.com/facebookresearch/AutoAvatar
๐AutoAvatar by #Meta: autoregressive method for modeling dynamically deforming human bodies from raw scans
๐Review https://bit.ly/3W0oTgo
๐Paper arxiv.org/pdf/2203.13817.pdf
๐Project zqbai-jeremy.github.io/autoavatar
๐Code github.com/facebookresearch/AutoAvatar
๐11๐ฅ2๐พ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฏModeling the Human Pose Manifolds๐ฏ
๐#Meta Pose-NDF: continuous model for plausible human poses based on neural distance fields
๐Review https://bit.ly/3f6X59o
๐Paper arxiv.org/pdf/2207.13807.pdf
๐Project virtualhumans.mpi-inf.mpg.de/posendf
๐Code github.com/garvita-tiwari/PoseNDF
๐#Meta Pose-NDF: continuous model for plausible human poses based on neural distance fields
๐Review https://bit.ly/3f6X59o
๐Paper arxiv.org/pdf/2207.13807.pdf
๐Project virtualhumans.mpi-inf.mpg.de/posendf
๐Code github.com/garvita-tiwari/PoseNDF
โค6๐คฏ4
This media is not supported in your browser
VIEW IN TELEGRAM
๐ Gesture Recognition in 80's ๐
๐The #Casio AT-550 was offering the edge gesture recognition in 1984!
๐Review https://bit.ly/3fcPia6
๐Clip: youtube.com/watch?v=piFaJmYpQfQ
๐The #Casio AT-550 was offering the edge gesture recognition in 1984!
๐Review https://bit.ly/3fcPia6
๐Clip: youtube.com/watch?v=piFaJmYpQfQ
๐คฏ15๐ฅฐ6๐ฅ2
This media is not supported in your browser
VIEW IN TELEGRAM
๐ธMUSIKA! Neural โ-audio generation๐ธ
๐Novel neural music generation system on single consumer GPU!
๐Listen: https://bit.ly/3W80p4U
๐Paper arxiv.org/pdf/2208.08706.pdf
๐Code github.com/marcoppasini/musika
๐Novel neural music generation system on single consumer GPU!
๐Listen: https://bit.ly/3W80p4U
๐Paper arxiv.org/pdf/2208.08706.pdf
๐Code github.com/marcoppasini/musika
๐คฏ8