- Accuracy 重建的模型精度不好，影响因素：
- 数据集质量：照片拍摄质量(设备)、相机位姿估计精度(COLMAP)
- 照片质量问题：混叠、模糊、滚动快门 (RS) 效应、HDR/LDR、运动模糊、低光照
- 相机位姿误差：SFM位姿估计时的误差
- NeuS方法的问题：体渲染公式的过度简化、表面几何与颜色的偏差、缺少几何约束(深度or法向量)
- 网格提取方法(Marching Cube)：分辨率太低
- 数据集质量：照片拍摄质量(设备)、相机位姿估计精度(COLMAP)
- Efficiency 训练/渲染的速度太慢，影响因素：
- MLP计算次数多 —> MIMO MLP、NGP-RT、
- MLP层数多计算慢—> InstantNGP
CAX
CAD/CAE/CAM
Vibration Measurement
一些振动测量的方法
Dimensions Measurement
如何从二维图片中测量三维物体的尺寸信息？
3D Model
三维模型的各种形式，初步探索
- 构造实体几何(eg: CAD)
- 多边形网格(eg: FEA)
- 近似网格编码——曲面细分
- 精确网格编码——样条(NURBS)
- 箱体建模(eg: 雕刻模型 for 游戏/电影)
Learn-Cpp
Learn-Rust
NeuralWarp
| Title | Improving neural implicit surfaces geometry with patch warping |
|---|---|
| Author | François Darmon1, 2 Bénédicte Bascle1 Jean-Clément Devaux1 Pascal Monasse2 Mathieu Aubry2 |
| Conf/Jour | 2022 |
| Year | CVPR |
| Project | Improving neural implicit surfaces geometry with patch warping |
| Paper | Improving neural implicit surfaces geometry with patch warping |
ActiveNeRF
| Title | ActiveNeRF |
|---|---|
| Author | Pan, Xuran and Lai, Zihang and Song, Shiji and Huang, Gao |
| Conf/Jour | ECCV |
| Year | 2022 |
| Project | LeapLabTHU/ActiveNeRF: Official repository of ActiveNeRF (ECCV2022) |
| Paper | ActiveNeRF: Learning where to See with Uncertainty Estimation |
Sources of Uncertainty in 3D Scene Reconstruction
| Title | Sources of Uncertainty in 3D Scene Reconstruction |
|---|---|
| Author | Marcus Klasson and Riccardo Mereu and Juho Kannala and Arno Solin |
| Conf/Jour | ECCV Workshop on Uncertainty Quantification for Computer Vision. |
| Year | 2024 |
| Project | AaltoML/uncertainty-nerf-gs: Code release for the paper “Sources of Uncertainty in 3D Scene Reconstruction” |
| Paper | Sources of Uncertainty in 3D Scene Reconstruction |
Dynamic Response Prediction
FE Surrogate Model:
How to get dynamic response (time domain) through structural parameters?
Structural Dynamics
Structural Dynamics Basic Knowledge.
mdof provides fast and friendly system identification for structures.
Data Analysis
数据降维/
sklearn | numpy | pandas
seaborn | matplotlib
Time Series Generation (Uncertainty)
Time Series Generation | Papers With Code
- 生成样本，解决样本不足的问题
Code of Multi-view 3D Reconstruction based on SDF and volume rendering
论文代码复现(Linux OS)
Learning-based Multi-View Stereo: A Survey
| Title | Learning-based Multi-View Stereo A Survey |
|---|---|
| Author | Fangjinhua Wang†, Qingtian Zhu, Di Chang*, Quankai Gao, Junlin Han, Tong Zhang, Richard Hartley, Fellow, IEEE, Marc Pollefeys, Fellow, IEEE |
| Conf/Jour | |
| Year | 2024 |
| Project | |
| Paper | Learning-based Multi-View Stereo: A Survey |
Visualize Interval uncertainty quantification metrics
使用Pygame可视化区间UQ指标
