Comments (4)
Hi, yes, it is possible to extend our current RGB-D implementation to work with a Stereo setup. Support for two images as input will require significant restructuring of the code. Hence, this is currently not on our radar.
from splatam.
Hi @Nik-V9 ,
I have a stereo data and used "https://wiki.ros.org/stereo_image_proc" ROS package to generate depth images.
I gave this data as input to the Splatam code.
I updated the code with necessary scripts like the data.yaml, config script, dataset class, etc. but the output received is not proper.
For the below scene (data is captured using the stereo setup mounted on a car. So the camera motion is in the forward direction only)
the output generated was
FRONT VIEW
But using Iphone 12 pro data, the output was proper.
The config files updated by me for above data are in below zip:
Splatam_updated_files.zip
Also wanted to understand how I can find "png_depth_scale" parameter for my data?
As far as my understanding I can use below formula:
where
- f is the focal length
- B is the baseline distance between the stereo cameras,
- D_image : is the pixel value in the depth image.
- Z_meters : Real world depth in meters.
Z_meters will be the png_depth_scale for my data. But this will not be constant, as Z_meters is inversely proportional to D_image.
Can you tell what I am doing wrong?
Any suggestions to fix my problem.
Thanks and Regards,
@darshanAtSwaayatt
from splatam.
Hi @Nik-V9 ,
I have a stereo data and used "https://wiki.ros.org/stereo_image_proc" ROS package to generate depth images. I gave this data as input to the Splatam code. I updated the code with necessary scripts like the data.yaml, config script, dataset class, etc. but the output received is not proper. For the below scene (data is captured using the stereo setup mounted on a car. So the camera motion is in the forward direction only)
the output generated was FRONT VIEW
But using Iphone 12 pro data, the output was proper.
The config files updated by me for above data are in below zip: Splatam_updated_files.zip
Also wanted to understand how I can find "png_depth_scale" parameter for my data? As far as my understanding I can use below formula:
where
- f is the focal length
- B is the baseline distance between the stereo cameras,
- D_image : is the pixel value in the depth image.
- Z_meters : Real world depth in meters.
Z_meters will be the png_depth_scale for my data. But this will not be constant, as Z_meters is inversely proportional to D_image.
Can you tell what I am doing wrong? Any suggestions to fix my problem.
Thanks and Regards, @darshanAtSwaayatt
Hi @Nik-V9 , Can you please help.
from splatam.
Hi, if the depth image you are saving has the pixel values in meters, the png_depth_scale
will be 1.
Essentially png_depth_scale
is a parameter to convert pixel values of a depth image to metric depth. In your case, after computing the metric depth if you are saving the image as is, the png_depth_scale
will be 1.
Also based on the stereo depth image shared by you, there seems to be a lot of holes in the depth map. Without good multi-view coverage, I'm unsure how well the current version of SplaTAM will perform on the data.
from splatam.
Related Issues (20)
- Question regarding the necessity of viewing cameras and relative pose transformations HOT 1
- Error installing rasterizer [warning: GCC] HOT 1
- Error when running SplaTAM on Scannet++ dataset HOT 2
- Rendering data with very near depth (< 0.2 m) [Duplicate] HOT 2
- Training with a live streaming HOT 1
- `RuntimeError: numel: integer multiplication overflow` HOT 3
- How to create a new dataset HOT 1
- Query regarding constant camera intrinsics HOT 8
- Type of Loss Function HOT 2
- GPU memory during training HOT 1
- Query regarding first camera pose HOT 2
- Applicable to Dynamic Avatar Reconstruction? HOT 2
- EGL error when trying to use your code to get the undistort depth from Scanet ++ HOT 2
- Discrepancies in Generated PLY File and Rendered Images HOT 1
- After running post_splatam_opt.py for densification, the newly generated npz file has worse performance compared to splatam.py HOT 4
- New Datasets with 3DGS HOT 1
- How to use ground truth camera pose instead of esimating one during reconstruction HOT 1
- Aligning 3D Gaussian with Original Point Cloud Coordinates
- Graphics card memory usage
- Query regarding Depth Backward Pass (Duplicate) HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
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.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from splatam.