QiYun

深度估计+SMPL 估计+Diffusion Model 精细化(PointNet++)

缺陷：依赖于深度估计方法和 SMPL 估计方法的精度
为了保护隐私不对人脸进行重建

No Code

Hello 算法 (hello-algo.com)

输入点云或低分辨率体素，提取特征后利用GAN网络，生成每个顶点的位置和SDF偏移值，得到优化后顶点的位置和SDF
结合显式与隐式表达的表示方法，利用MT，从隐式SDF中重建出显式mesh

PointCloud

• Registration
• Surface Reconstruction

Follow

• Yochengliu/awesome-point-cloud-analysis: A list of papers and datasets about point cloud analysis (processing) (github.com)
• zhulf0804/3D-PointCloud: Papers and Datasets about Point Cloud. (github.com)
• XuyangBai/awesome-point-cloud-registration: A curated list of point cloud registration. (github.com)

• 关注建筑物的重建，为 3D Diffusion Models 中添加了图片的信息嵌入(预训练了一个图片编码器)
• 两阶段的点云降噪模型，第一阶段关注全局，第二阶段关注细节
• 提出了两个自定义的新数据集，并在数据集上验证了本方法的优点

Stereoscopic-aware graph generative adversarial network (SG-GAN)

• flow-based explicit generative models for sampling point clouds with arbitrary resolutions
• Improving the detailed 3D structures of point clouds by leveraging the implicit generative adversarial networks (GANs).

New 3D Diffusion Transformer Model, 在体素化的点云上运行 DDPM(Denoising diffusion probabilistic models) 的去噪过程

按步长穷举法，粗配准，需要根据ICP进行精配准

pairwise registration models, only Ubuntu

Issue:
How should I train my dataset? · Issue #4 · WHU-USI3DV/SGHR (github.com)
I think several point clouds of a single statue is not enough for training deep descriptors. I suggest to directly use pairwise registration models such as Geotrainsformer pre-trained on object-level datasets such as ModelNet40 to solve the pairwise registrations.
And adopt SGHR’s transformation synchronization section to solve the global consistent scan poses.

Math basic of 3D Reconstruction

Terminology/Jargon

• Human Radiance Fields
• 3D Clothed Human Reconstruction | Digitization

Application

• 三维重建设备：手持扫描仪或 360 度相机矩阵（成本高）
• 复刻一个迷你版的自己

Method

1. Depth&Normal Estimation(2K2K)
2. Implicit Function(PIFu or NeRF)
3. Generative approach Generative Models Reconstruction

Awesome Human Body Reconstruction