๐ฅ๐ญ๐ New Research Alert - HeadGAP (Avatars Collection)! ๐๐ญ๐ฅ ๐ Title: HeadGAP: Few-shot 3D Head Avatar via Generalizable Gaussian Priors ๐
๐ Description: HeadGAP introduces a novel method for generating high-fidelity, animatable 3D head avatars from few-shot data, using Gaussian priors and dynamic part-based modelling for personalized and generalizable results.
๐๐บ๐ New Research Alert - ECCV 2024 (Avatars Collection)! ๐๐๐ ๐ Title: Expressive Whole-Body 3D Gaussian Avatar ๐
๐ Description: ExAvatar is a model that generates animatable 3D human avatars with facial expressions and hand movements from short monocular videos using a hybrid mesh and 3D Gaussian representation.
๐ฅ Authors: Gyeongsik Moon, Takaaki Shiratori, and @psyth
๐ฅ๐ญ๐ New Research Alert - ECCV 2024 (Avatars Collection)! ๐๐ญ๐ฅ ๐ Title: MeshAvatar: Learning High-quality Triangular Human Avatars from Multi-view Videos ๐
๐ Description: MeshAvatar is a novel pipeline that generates high-quality triangular human avatars from multi-view videos, enabling realistic editing and rendering through a mesh-based approach with physics-based decomposition.
๐ฅ Authors: Yushuo Chen, Zerong Zheng, Zhe Li, Chao Xu, and Yebin Liu
๐๐บ๐ New Research Alert - CVPR 2024 (Avatars Collection)! ๐๐๐ ๐ Title: IntrinsicAvatar: Physically Based Inverse Rendering of Dynamic Humans from Monocular Videos via Explicit Ray Tracing ๐
๐ Description: IntrinsicAvatar is a method for extracting high-quality geometry, albedo, material, and lighting properties of clothed human avatars from monocular videos using explicit ray tracing and volumetric scattering, enabling realistic animations under varying lighting conditions.
๐ฅ Authors: Shaofei Wang, Boลพidar Antiฤ, Andreas Geiger, and Siyu Tang
๐ Conference: CVPR, Jun 17-21, 2024 | Seattle WA, USA ๐บ๐ธ
๐ฅ๐ญ๐ New Research Alert - ECCV 2024 (Avatars Collection)! ๐๐ญ๐ฅ ๐ Title: RodinHD: High-Fidelity 3D Avatar Generation with Diffusion Models ๐
๐ Description: RodinHD generates high-fidelity 3D avatars from portrait images using a novel data scheduling strategy and weight consolidation regularization to capture intricate details such as hairstyles.
๐ฅ๐ญ๐ New Research Alert - LivePortrait (Avatars Collection)! ๐๐ญ๐ฅ ๐ Title: LivePortrait: Efficient Portrait Animation with Stitching and Retargeting Control ๐
๐ Description: LivePortrait is an efficient video-driven portrait animation framework that uses implicit keypoints and stitching/retargeting modules to generate high-quality, controllable animations from a single source image.
๐ฅ Authors: @cleardusk, Dingyun Zhang, Xiaoqiang Liu, Zhizhou Zhong, Yuan Zhang, Pengfei Wan, and Di Zhang
๐๐บ๐ New Research Alert (Avatars Collection)! ๐๐๐ ๐ Title: Expressive Gaussian Human Avatars from Monocular RGB Video ๐
๐ Description: The new EVA model enhances the expressiveness of digital avatars by using 3D Gaussians and SMPL-X to capture fine-grained hand and face details from monocular RGB video.
๐ฅ Authors: Hezhen Hu, Zhiwen Fan, Tianhao Wu, Yihan Xi, Seoyoung Lee, Georgios Pavlakos, and Zhangyang Wang
๐ฅ๐ญ๐ New Research Alert - ECCV 2024 (Avatars Collection)! ๐๐ญ๐ฅ ๐ Title: Topo4D: Topology-Preserving Gaussian Splatting for High-Fidelity 4D Head Capture ๐
๐ Description: Topo4D is a novel method for automated, high-fidelity 4D head tracking that optimizes dynamic topological meshes and 8K texture maps from multi-view time-series images.
๐ฅ Authors: @Dazz1e, Y. Cheng, @Ryan-sjtu, H. Jia, D. Xu, W. Zhu, Y. Yan
๐๐ญ๐ New Research Alert - Portrait4D-v2 (Avatars Collection)! ๐๐ญ๐ ๐ Title: Portrait4D-v2: Pseudo Multi-View Data Creates Better 4D Head Synthesizer ๐
๐ Description: Portrait4D-v2 is a novel method for one-shot 4D head avatar synthesis using pseudo multi-view videos and a vision transformer backbone, achieving superior performance without relying on 3DMM reconstruction.
๐ฅ Authors: Yu Deng, Duomin Wang, and Baoyuan Wang
๐๐ฒ๐๐ก New Research Alert - CVPRW 2024 (Facial Expressions Recognition Collection)! ๐ก๐ฅ๐ฅด๐ฑ ๐ Title: Zero-Shot Audio-Visual Compound Expression Recognition Method based on Emotion Probability Fusion ๐
๐ Description: AVCER is a novel audio-visual method for compound expression recognition based on pair-wise sum of emotion probability, evaluated in multi- and cross-corpus setups without task-specific training data, demonstrating its potential for intelligent emotion annotation tools.
๐๐ญ๐ New Research Alert - CVPR 2024 (Avatars Collection)! ๐๐ญ๐ ๐ Title: Relightable Gaussian Codec Avatars ๐
๐ Description: Relightable Gaussian Codec Avatars is a method for creating highly detailed and relightable 3D head avatars that can animate expressions in real time and support complex features such as hair and skin with efficient rendering suitable for VR.
๐๐ญ๐ New Research Alert - InstructAvatar (Avatars Collection)! ๐๐ญ๐ ๐ Title: InstructAvatar: Text-Guided Emotion and Motion Control for Avatar Generation ๐
๐ Description: InstructAvatar is a novel method for generating emotionally expressive 2D avatars using text-guided instructions, offering improved emotion control, lip-sync quality, and naturalness. It uses a two-branch diffusion-based generator to predict avatars based on both audio and text input.
๐ฅ๐๐ New Research Alert - YOLOv10! ๐๐๐ฅ ๐ Title: YOLOv10: Real-Time End-to-End Object Detection ๐
๐ Description: YOLOv10 improves real-time object recognition by eliminating non-maximum suppression and optimizing the model architecture to achieve state-of-the-art performance with lower latency and computational overhead.
๐๐ญ๐ New Research Alert - Gaussian Head & Shoulders (Avatars Collection)! ๐๐ญ๐ ๐ Title: Gaussian Head & Shoulders: High Fidelity Neural Upper Body Avatars with Anchor Gaussian Guided Texture Warping ๐
๐ Description: Gaussian Head & Shoulders is a method for creating high-fidelity upper body avatars by integrating 3D morphable head models with a neural texture warping approach to overcome the limitations of Gaussian splatting.
๐๐ค๐ New Research Alert - CVPR 2024! ๐๐ค๐ ๐ Title: RoHM: Robust Human Motion Reconstruction via Diffusion ๐
๐ Description: RoHM is a diffusion-based approach for robust 3D human motion reconstruction from monocular RGB(-D) videos, effectively handling noise and occlusions to produce complete and coherent motions. This method outperforms current techniques in various tasks and is faster at test time.
๐ฅ Authors: Siwei Zhang et al.
๐ Conference: CVPR, Jun 17-21, 2024 | Seattle WA, USA ๐บ๐ธ
๐๐๐ New Research Alert - SIGGRAPH 2024 (Avatars Collection)! ๐๐๐ ๐ Title: LayGA: Layered Gaussian Avatars for Animatable Clothing Transfer ๐
๐ Description: LayGA is a novel method for animatable clothing transfer that separates the body and clothing into two layers for improved photorealism and accurate clothing tracking, outperforming existing methods.
๐ฅ Authors: Siyou Lin, Zhe Li, Zhaoqi Su, Zerong Zheng, Hongwen Zhang, and Yebin Liu
