GithubHelp home page GithubHelp logo

dvms's Introduction

Datasets

Wu_MMSys_17: 即vr-dataset; https://dl.acm.org/doi/10.1145/3083187.3083210

Nasrabadi_MMSys_19: https://dl.acm.org/doi/abs/10.1145/3304109.3325812

David_MMSys_18: https://dl.acm.org/doi/abs/10.1145/3204949.3208139

NOTE:

  • A-large-dataset 的6个数据集中, 有4个数据集未在论文中明确 "视点位置坐标系" 与 "ERP格式的视频画面" 在位置上的对应关系, 因而无法进行对齐. 这里选择剩余两个 ([4] (Wu_MMSys_17) 和 [6] (Nasrabadi_MMSys_19) )进行处理.
  • David_MMSys_18数据集没有包含在A-large-dataset中, 是我之前做视点预测一直在用的主数据集, 因为它体积小, 训练测试块, 而且视点运动幅度大, HT+VC的VP模型更容易取得比较好的性能.

Directory structure

DVMS
├── Wu_MMSys_17
│   ├── dataset
│   ├── sampled_dataset_thetaphi
│   │   ├── video1
│   │   │   ├── user1
│   │   │   ├── user2
│   │   │   └── ...
│   │   ├── video2
│   │   └── ...
│   └── saliency_src
│       ├── video1.npy
│       ├── video2.npy
│       └── ...
├── Nasrabadi_MMSys_19
└── David_MMSys_18
  • dataset : 原始数据集
  • sampled_dataset_thetaphi : 师兄最终需要的视点数据;
    • 共有3列: ['playback_time(s)', 'theta', 'phi' ], 其中: theta ranges from 0 to 2*pi, and phi ranges from 0 to pi, origin of equirectangular in the top-left corner;
    • 采样率: 5Hz; 即0.2s采样一次;
    • 每个数据文件的读取方式为: data = pd.read_csv(path, header=None);
  • saliency_src :
    • shape = (采样点数量, 224, 448); 即原始的saliency map的形状为224x448;
    • 可通过运行项目根目录下的 resize_salmaps.py 脚本将 saliency_src resize 成任意尺寸;

Pipeline (using dataset Wu_MMSys_17 as an example)

cd path/to/DVMS
conda activate gbq_pytorch 

Head movement traces

python Wu_MMSys_17/Read_Dataset.py --creat_orig_dat  # dataset --> original_dataset_xyz (统一目录结构, 以及视点位置的表示格式 (x,y,z) )
python Wu_MMSys_17/Read_Dataset.py --creat_samp_dat  # original_dataset_xyz --> sampled_dataset (统一采样率为5Hz, 即0.2s一个数据点)
python Wu_MMSys_17/Read_Dataset.py --creat_thph_dat  # sampled_dataset --> sampled_dataset_thetaphi (将视点位置表示格式从(x,y,z)转成(theta, phi) ; theta ranges from 0 to 2*pi, and phi ranges from 0 to pi, origin of equirectangular in the top-left corner)

trivial: 压缩以方便传输:

cd Wu_MMSys_17
zip -r dataset.zip ./dataset/
zip -r sampled.zip ./sampled_dataset_thetaphi/

Saliency maps

生成:

https://github.com/bzsgbq/PAVER-main

resize:

  • 首先打开项目根目录下的 resize_salmaps.py, 配置dataset_name和SALMAP_SHAPE;
  • 然后直接在项目根目录下运行脚本: python resize_salmaps.py

Main References

https://gitlab.com/miguelfromeror/head-motion-prediction

https://gitlab.com/DVMS_/DVMS

https://github.com/HS-YN/PAVER

dvms's People

Contributors

bzsgbq avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.