整理：AI算法与图像处理
CVPR2022论文和代码整理：https://github.com/DWCTOD/CVPR2022-Papers-with-Code-Demo
ECCV2022论文和代码整理：https://github.com/DWCTOD/ECCV2022-Papers-with-Code-Demo
最新成果demo展示：
ECCV2022 | 帅，太帅了！联合多目标检测、3D 纹理重建、6D 目标姿态和大小估计的方法
论文：https://arxiv.org/abs/2207.13691
代码：https://zubair-irshad.github.io/projects/ShAPO.html
ECCV2022 汇总：https://github.com/DWCTOD/ECCV2022-Papers-with-Code-Demo/
摘要：
我们的方法从单个 RGB-D 观察中研究以对象为中心的整体 3D 理解的复杂任务。由于这是一个不适定问题，现有方法在具有遮挡的复杂多对象场景中的 3D 形状和 6D 姿态估计性能低下。我们提出了 ShaAPO，一种用于联合多对象检测、3D 纹理重建、6D 对象姿态和大小估计的方法。ShAPO 的关键是一个单次管道，用于回归形状、外观和姿势潜在代码以及每个对象实例的掩码，然后以稀疏到密集的方式进一步细化。首先学习一种新颖的解开形状和外观代码，将对象嵌入到它们各自的形状和外观空间中。我们还提出了一种新颖的、基于八叉树的可微优化步骤，使我们能够以综合分析的方式在学习的潜在空间下同时进一步改进对象形状、姿势和外观。我们新颖的联合隐式纹理对象表示使我们能够准确地识别和重建新的看不见的对象，而无需访问它们的 3D 网格。通过广泛的实验，我们证明了我们的方法在模拟室内场景上进行训练，能够以最少的微调准确地回归现实世界中新物体的形状、外观和姿势。
最新论文整理
ECCV2022
Updated on : 29 Jul 2022
total number : 12
Initialization and Alignment for Adversarial Texture Optimization
- 论文/Paper: http://arxiv.org/pdf/2207.14289
- 代码/Code: None
Depth Field Networks for Generalizable Multi-view Scene Representation
- 论文/Paper: http://arxiv.org/pdf/2207.14287
- 代码/Code: None
The One Where They Reconstructed 3D Humans and Environments in TV Shows
- 论文/Paper: http://arxiv.org/pdf/2207.14279
- 代码/Code: None
MonteBoxFinder: Detecting and Filtering Primitives to Fit a Noisy Point Cloud
- 论文/Paper: http://arxiv.org/pdf/2207.14268
- 代码/Code: https://github.com/MichaelRamamonjisoa/MonteBoxFinder
Mining Cross-Person Cues for Body-Part Interactiveness Learning in HOI Detection
- 论文/Paper: http://arxiv.org/pdf/2207.14192
- 代码/Code: https://github.com/enlighten0707/Body-Part-Map-for-Interactiveness.
Neural Strands: Learning Hair Geometry and Appearance from Multi-View Images
- 论文/Paper: http://arxiv.org/pdf/2207.14067
- 代码/Code: None
Video Mask Transfiner for High-Quality Video Instance Segmentation
- 论文/Paper: http://arxiv.org/pdf/2207.14012
- 代码/Code: None
Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers
- 论文/Paper: http://arxiv.org/pdf/2207.13820
- 代码/Code: https://github.com/postech-ami/FastMETRO
Look at Adjacent Frames: Video Anomaly Detection without Offline Training
- 论文/Paper: http://arxiv.org/pdf/2207.13798
- 代码/Code: None
AvatarPoser: Articulated Full-Body Pose Tracking from Sparse Motion Sensing
- 论文/Paper: http://arxiv.org/pdf/2207.13784
- 代码/Code: https://github.com/eth-siplab/AvatarPoser
Break and Make: Interactive Structural Understanding Using LEGO Bricks
- 论文/Paper: http://arxiv.org/pdf/2207.13738
- 代码/Code: https://github.com/aaronwalsman/ltron.
A Repulsive Force Unit for Garment Collision Handling in Neural Networks
- 论文/Paper: http://arxiv.org/pdf/2207.13871
- 代码/Code: None
CVPR2022
Updated on : 29 Jul 2022
total number : 1
PEA: Improving the Performance of ReLU Networks for Free by Using Progressive Ensemble Activations
- 论文/Paper: http://arxiv.org/pdf/2207.14074
- 代码/Code: None
