dazinovic / neural-rgbd-surface-reconstruction

Official implementation of the CVPR 2022 Paper "Neural RGB-D Surface Reconstruction"

Home Page:https://dazinovic.github.io/neural-rgbd-surface-reconstruction/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

How to make own datasets?

divingwolf opened this issue · comments

Hello, author.

Thank you for completing and providing this great work.

I ran the code and got perfect results on the data set you provided, so I want to try to make my own data set and see how the code works. Could you describe in detail how to create own datasets, such as how to use BundleFusion you mentioned in readme.md.

Hello, @jinshuitiaofan may I ask how did you prepare the data folder? I have downloaded the datasets but I have no idea how to prepare the files listed in data section such as:
├── depth0.png
├── img1.png
├── focal.txt
├── poses.txt
├── trainval_poses.txt

Hello, @jinshuitiaofan may I ask how did you prepare the data folder? I have downloaded the datasets but I have no idea how to prepare the files listed in data section such as: ├── depth0.png ├── img1.png ├── focal.txt ├── poses.txt ├── trainval_poses.txt

haha, I dont know. The author has not replied to me yet.

I think I figured it out thanks all:)

I think I figured it out thanks all:)

bro, show me your achievement. :)

Hi! You should organize it in the following way, as README.md says:

<scene_name>            # datadir in the config file
├── depth
    ├── leave empty
    ...
├── depth_filtered      # your depth images
    ├── depth0.png     
    ├── depth1.png
    ├── depth2.png
    ...
├── images              # your RGB images
    ├── img0.png     
    ├── img1.png
    ├── img2.png
    ...
├── focal.txt           # your camera focal length
├── trainval_poses.txt  # your camera poses for optimization