AI with Papers - Artificial Intelligence & Deep Learning
๐ Drag-GAN: user-friendly image-manipulation ๐ ๐ Manual deforming of (real and generated) images over pose, shape, expression and layout. ๐Review https://bit.ly/3BFyXlR ๐Paper arxiv.org/pdf/2305.10973.pdf ๐Project vcai.mpi-inf.mpg.de/projects/DragGANโฆ
Linkedin
๐ฅ๐ฅ Source Code of Drag-GAN IS OUT! | Alessandro Ferrari | 40 comments
๐ฅ๐ฅ Source Code of Drag-GAN IS OUT! ๐ฅ๐ฅ
๐Manual deforming of (real and generated) images over pose, shape, expression and layout. Source Code just released a few hours ago ๐
๐๐ข๐ ๐ก๐ฅ๐ข๐ ๐ก๐ญ๐ฌ:
โ Max Planck + MIT + #Google AR/VR = ๐คฏ
โ Supervising handle points to moveโฆ
๐Manual deforming of (real and generated) images over pose, shape, expression and layout. Source Code just released a few hours ago ๐
๐๐ข๐ ๐ก๐ฅ๐ข๐ ๐ก๐ญ๐ฌ:
โ Max Planck + MIT + #Google AR/VR = ๐คฏ
โ Supervising handle points to moveโฆ
๐ฅ25๐ฑ6โค3๐ฅฐ1๐คฏ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฎSAM-PT: Segment Anything+Tracking๐ฎ
๐SAM-PT is the first method to utilize sparse point propagation for Video Object Segmentation (VOS).
๐Review https://t.ly/QLMG
๐Paper arxiv.org/pdf/2307.01197.pdf
๐Project www.vis.xyz/pub/sam-pt/
๐Code github.com/SysCV/sam-pt
๐SAM-PT is the first method to utilize sparse point propagation for Video Object Segmentation (VOS).
๐Review https://t.ly/QLMG
๐Paper arxiv.org/pdf/2307.01197.pdf
๐Project www.vis.xyz/pub/sam-pt/
๐Code github.com/SysCV/sam-pt
๐ฅ14โค7๐คฏ3๐1๐ฑ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ชฉ DISCO: Human Dance Generation ๐ชฉ
๐NTU (+ #Microsoft) unveils DISCO: a big step towards the Human Dance Generation.
๐Review https://t.ly/cNGX
๐Paper arxiv.org/pdf/2307.00040.pdf
๐Project disco-dance.github.io/
๐Code github.com/Wangt-CN/DisCo
๐NTU (+ #Microsoft) unveils DISCO: a big step towards the Human Dance Generation.
๐Review https://t.ly/cNGX
๐Paper arxiv.org/pdf/2307.00040.pdf
๐Project disco-dance.github.io/
๐Code github.com/Wangt-CN/DisCo
๐ฅ13๐ฅฐ4๐2โก1๐1๐พ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฃ๏ธ STAR.: 3D-tracking w/ attention paradigm ๐ฃ๏ธ
๐#Mercedes STAR: e2e 3D object tracking that follows the tracking-by-attention paradigm
๐Review https://t.ly/JoGj
๐Paper arxiv.org/pdf/2306.17602.pdf
๐Project simondoll.github.io/publications/star_track
๐#Mercedes STAR: e2e 3D object tracking that follows the tracking-by-attention paradigm
๐Review https://t.ly/JoGj
๐Paper arxiv.org/pdf/2306.17602.pdf
๐Project simondoll.github.io/publications/star_track
๐14๐ฅ1๐ฅฐ1๐1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ก Text2Cinemagraphs: Cinemagraph from text ๐ก
๐CMU (+ #Snap) unveils a fully automated method for creating cinemagraphs from text descriptions
๐Review https://t.ly/BwZs6
๐Paper arxiv.org/pdf/2307.03190.pdf
๐Project text2cinemagraph.github.io/website
๐Code github.com/text2cinemagraph/text2cinemagraph
๐CMU (+ #Snap) unveils a fully automated method for creating cinemagraphs from text descriptions
๐Review https://t.ly/BwZs6
๐Paper arxiv.org/pdf/2307.03190.pdf
๐Project text2cinemagraph.github.io/website
๐Code github.com/text2cinemagraph/text2cinemagraph
โค12๐คฏ3๐ฑ1๐คฉ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅTest-Time Training on fire ๐ฅ
๐Extending the TTT to the streaming setting. Suitable for Panoptic, Instance & Colorization.
๐Review https://t.ly/eZYA
๐Paper arxiv.org/pdf/2307.05014.pdf
๐Project https://video-ttt.github.io/
๐Code github.com/renwang435/video-ttt-release
๐Extending the TTT to the streaming setting. Suitable for Panoptic, Instance & Colorization.
๐Review https://t.ly/eZYA
๐Paper arxiv.org/pdf/2307.05014.pdf
๐Project https://video-ttt.github.io/
๐Code github.com/renwang435/video-ttt-release
๐ฅ10๐3โก1๐คฏ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ Deepfake via casual self-scan ๐
๐TAU presents a novel approach to reenact an ID using only a casual self-scan
๐Review https://t.ly/9T8Wi
๐Paper arxiv.org/pdf/2307.06307.pdf
๐Project arielazary.github.io/PGR
๐TAU presents a novel approach to reenact an ID using only a casual self-scan
๐Review https://t.ly/9T8Wi
๐Paper arxiv.org/pdf/2307.06307.pdf
๐Project arielazary.github.io/PGR
๐คฏ7๐6โค5๐ฅ1๐ฑ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ช Extreme Human Pose Estimation ๐ช
๐RePoGen: novel synthetic data generator of extreme/realistic poses of humans
๐Review https://t.ly/ecBvM
๐Paper arxiv.org/pdf/2307.06737.pdf
๐Project mirapurkrabek.github.io/RePoGen-paper
๐Code github.com/MiraPurkrabek/RePoGen
๐RePoGen: novel synthetic data generator of extreme/realistic poses of humans
๐Review https://t.ly/ecBvM
๐Paper arxiv.org/pdf/2307.06737.pdf
๐Project mirapurkrabek.github.io/RePoGen-paper
๐Code github.com/MiraPurkrabek/RePoGen
๐ฅ12๐2๐1๐คฏ1๐ฑ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ก DATID-3D: Text-to-3D Generation ๐ก
๐ A novel domain adaptation method for 3D via text-to-image diffusion. ๐ค-Demo available!
๐Review https://t.ly/TCL-B
๐Paper arxiv.org/pdf/2211.16374.pdf
๐Project gwang-kim.github.io/datid_3d/
๐Code github.com/gwang-kim/DATID-3D
๐ค huggingface.co/spaces/gwang-kim/DATID-3D
๐Colab colab.research.google.com/drive/1e9NSVB7x_hjz-nr4K0jO4rfTXILnNGtA?usp=sharing
๐ A novel domain adaptation method for 3D via text-to-image diffusion. ๐ค-Demo available!
๐Review https://t.ly/TCL-B
๐Paper arxiv.org/pdf/2211.16374.pdf
๐Project gwang-kim.github.io/datid_3d/
๐Code github.com/gwang-kim/DATID-3D
๐ค huggingface.co/spaces/gwang-kim/DATID-3D
๐Colab colab.research.google.com/drive/1e9NSVB7x_hjz-nr4K0jO4rfTXILnNGtA?usp=sharing
๐คฏ5
This media is not supported in your browser
VIEW IN TELEGRAM
๐งฏNeural Focal Modulation VAR๐งฏ
๐A novel architecture for video recognition that models both local/global context
๐Review https://t.ly/rF_fk
๐Paper arxiv.org/pdf/2307.06947.pdf
๐Project talalwasim.github.io/Video-FocalNets
๐Code github.com/TalalWasim/Video-FocalNets
๐A novel architecture for video recognition that models both local/global context
๐Review https://t.ly/rF_fk
๐Paper arxiv.org/pdf/2307.06947.pdf
๐Project talalwasim.github.io/Video-FocalNets
๐Code github.com/TalalWasim/Video-FocalNets
๐ฅ8โก1๐1๐คฉ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ Gen-AI as representation learner ๐
๐DreamTeacher: novel self-supervised feats. representation learning framework that utilizes gen-nets for pre-training downstream image backbones
๐Review https://t.ly/RL8iG
๐Paper arxiv.org/pdf/2307.07487.pdf
๐Project research.nvidia.com/labs/toronto-ai/DreamTeacher
๐DreamTeacher: novel self-supervised feats. representation learning framework that utilizes gen-nets for pre-training downstream image backbones
๐Review https://t.ly/RL8iG
๐Paper arxiv.org/pdf/2307.07487.pdf
๐Project research.nvidia.com/labs/toronto-ai/DreamTeacher
๐ฅ9๐2๐คฏ1
This media is not supported in your browser
VIEW IN TELEGRAM
โ #SelfDriving? It's all about weather! โ
๐Novel self-supervised MDE method to handle adverse weather in real-world autonomous driving
๐Review https://t.ly/tcLQW
๐Paper arxiv.org/pdf/2307.08357.pdf
๐Project kieran514.github.io/Robust-Depth-Project/
๐Novel self-supervised MDE method to handle adverse weather in real-world autonomous driving
๐Review https://t.ly/tcLQW
๐Paper arxiv.org/pdf/2307.08357.pdf
๐Project kieran514.github.io/Robust-Depth-Project/
โค7๐3๐คฏ1๐ฑ1
๐ฆ Llama-2: the Open-Source "ChatGPT" ๐ฆ
๐GenAI, #Meta unveils Llama-2: a collection of LLMs ranging in scale 7-70B params. Challenging with #chatgpt, but open.
๐Review https://t.ly/bLJgP
๐Paper https://t.ly/AOXru
๐Project https://ai.meta.com/llama
๐GenAI, #Meta unveils Llama-2: a collection of LLMs ranging in scale 7-70B params. Challenging with #chatgpt, but open.
๐Review https://t.ly/bLJgP
๐Paper https://t.ly/AOXru
๐Project https://ai.meta.com/llama
๐คฏ19โค2๐ฅ1๐ฉ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ AltFreezing: new SOTA in detecting deepfake ๐
๐#Microsoft unveils AltFreezing: spatial/temporal artifacts in one model for more general face forgery detection
๐Review https://t.ly/mkIKX
๐Paper https://t.ly/z4KnJ
๐Code github.com/ZhendongWang6/AltFreezing
๐#Microsoft unveils AltFreezing: spatial/temporal artifacts in one model for more general face forgery detection
๐Review https://t.ly/mkIKX
๐Paper https://t.ly/z4KnJ
๐Code github.com/ZhendongWang6/AltFreezing
๐ฑ6๐5๐4๐คฏ2๐ฅฐ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ชMETA's Ultra-HD Data for #AR๐ช
๐Aria Digital Twin: egocentric dataset for detection/tracking, reconstruction/understanding, S2R learning, pose and more.
๐Review https://t.ly/MRPt1
๐Paper arxiv.org/pdf/2306.06362.pdf
๐Project www.projectaria.com/datasets/adt
๐Code github.com/facebookresearch/projectaria_tools
๐Aria Digital Twin: egocentric dataset for detection/tracking, reconstruction/understanding, S2R learning, pose and more.
๐Review https://t.ly/MRPt1
๐Paper arxiv.org/pdf/2306.06362.pdf
๐Project www.projectaria.com/datasets/adt
๐Code github.com/facebookresearch/projectaria_tools
๐ฅ10๐1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฉโ๐ฆฐ Ultra-Realistic Neural Hair ๐ฉโ๐ฆฐ
๐A novel method to reconstruct the hair geometry at a strand level from monocular video or multi-view images
๐Review https://t.ly/6xZyp
๐Paper arxiv.org/pdf/2306.05872.pdf
๐Project samsunglabs.github.io/NeuralHaircut
๐Code github.com/SamsungLabs/NeuralHaircut
๐A novel method to reconstruct the hair geometry at a strand level from monocular video or multi-view images
๐Review https://t.ly/6xZyp
๐Paper arxiv.org/pdf/2306.05872.pdf
๐Project samsunglabs.github.io/NeuralHaircut
๐Code github.com/SamsungLabs/NeuralHaircut
๐คฏ17๐คฉ5๐5๐2โก1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ช Muscles in Action with #AI ๐ช
๐Muscles in Action (MIA): learn to incorporate muscle activity into human motion representations
๐Review https://t.ly/hUKub
๐Paper arxiv.org/pdf/2212.02978.pdf
๐Project musclesinaction.cs.columbia.edu
๐Muscles in Action (MIA): learn to incorporate muscle activity into human motion representations
๐Review https://t.ly/hUKub
๐Paper arxiv.org/pdf/2212.02978.pdf
๐Project musclesinaction.cs.columbia.edu
๐ฅ7๐2๐2๐คฏ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ชค PAPR: Proximity Attention Point Render ๐ชค
๐PAPR: fast point-based scene representation with differentiable renderer approach
๐Review https://t.ly/yoI0g
๐Paper arxiv.org/pdf/2307.11086.pdf
๐Project https://zvict.github.io/papr
๐PAPR: fast point-based scene representation with differentiable renderer approach
๐Review https://t.ly/yoI0g
๐Paper arxiv.org/pdf/2307.11086.pdf
๐Project https://zvict.github.io/papr
๐2๐ฅฐ2๐ฑ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ช CAD-based Object Segmentation ๐ช
๐ A novel three-stage approach to segment unseen objects in RGB images using their CAD models
๐Review https://t.ly/RtHLN
๐Paper arxiv.org/pdf/2307.11067.pdf
๐Code https://github.com/nv-nguyen/cnos
๐ A novel three-stage approach to segment unseen objects in RGB images using their CAD models
๐Review https://t.ly/RtHLN
๐Paper arxiv.org/pdf/2307.11067.pdf
๐Code https://github.com/nv-nguyen/cnos
๐ฅ7๐คฏ4โค1๐ฑ1๐คฉ1๐1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ต ALPR via CTS-Matching ๐ต
๐UIT unveils a neural approach (#YOLO5 + tracking + rotation) to improve the license plate recognition accuracy
๐Review https://t.ly/VP4BP
๐Paper arxiv.org/pdf/2307.11336.pdf
๐Code github.com/chequanghuy/Character-Time-series-Matching
๐UIT unveils a neural approach (#YOLO5 + tracking + rotation) to improve the license plate recognition accuracy
๐Review https://t.ly/VP4BP
๐Paper arxiv.org/pdf/2307.11336.pdf
๐Code github.com/chequanghuy/Character-Time-series-Matching
๐ฅ9โค2๐คฏ1๐ฑ1๐คฃ1
This media is not supported in your browser
VIEW IN TELEGRAM
๐ฅฌ Generative AIโs Next Frontiers ๐ฅฌ
๐Hair simulation, 2D->3D animation, and much more. ~20 papers from #NVIDIA accepted into #SIGGRAPH2023
๐ Review https://t.ly/wgGin
๐Hair simulation, 2D->3D animation, and much more. ~20 papers from #NVIDIA accepted into #SIGGRAPH2023
๐ Review https://t.ly/wgGin
๐คฏ13๐3๐คฉ3๐ฅฐ1๐ฑ1๐ฉ1