Coder Social home page Coder Social logo

shubhtuls / volumetricprimitives Goto Github PK

View Code? Open in Web Editor NEW
147.0 9.0 34.0 7.24 MB

Code release for "Learning Shape Abstractions by Assembling Volumetric Primitives " (CVPR 2017)

Lua 73.12% MATLAB 22.66% C 3.93% Shell 0.10% Python 0.19%
deep-learning unsupervised-learning abstraction 3d-representation

volumetricprimitives's People

Contributors

shubhtuls avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

volumetricprimitives's Issues

pretrained model

Hi!
Thank you for sharing the code. I am tring to running the demo code, but I found the link for downloading the pretrained model is broken. I wonder could you fix the link?
Thanks

how can I visualize the results?

Can you please let me know how can I visualize the results as you display in the project page. I mean the code that is used for rendering out the results.
Thanks

Dataset

Hi Shubhtuls,

After much investigation in the dataset, I've found some interesting properties in each mat files. They are tsdf and closest points. But in the paper, these are only needed when calculating the loss. Can you provide more information about them?

Thanks

Error while running precomputeShapeData

Hi,
I am getting the following error on running precomputeShapeData file.
error: 'extractfield' undefined near line 34 column 34
error: called from
getFileNamesFromDirectory at line 34 column 20
precomputeShapeData at line 18 column 16

About the initialization of the primitives

Thank you for your perfect work. I have a question about the initialization of the primitives. Could please show me where you initialize the position, shape and transformation of the primitives in your code? Thank you very much.

Question: how do you bias the cuboids to be small?

Hello,

First of all thanks for sharing the code. I have a couple of questions, how do you bias the cuboids to be small in the first step of the learning? Do you clip the output of the network? Also, how does the last layer that predicts the size, probability, translation and rotation of the primitive looks like? What activation functions have you used? Can you point me where can I find it in the code?

Thank you for your help!

Loss

local tsdfComputerTrain = nn.Sequential():add(transformer.partComposition(pgenFuncTsdf, params.nParts, nSamplePointsTrain)):add(nn.ReLU()):add(nn.Power(0.5*params.lossPower))

Hi Shubham,

Can you please kindly explain how is the loss defined on this line related to the consistency loss and coverage loss defined in the paper?

Thanks,
Shumin

What does "Tsdf" in the code stands for?

Hi Shubham,

I'm an undergrad student working on shape primitives decomposition and found the "Learning Shape Abstractions by Assembling Volumetric Primitives" paper, your project extremely interesting.

But I'm quite confused by the acronym "Tsdf". I'm guessing "df" stands for "distance field" but couldn't come up with a good hypothesis for "Ts" after much thinking (is it transformed?).

Could you please hint me at what they are referring to? It'd be really helpful.

Best,
Yichao

not found : point_mesh_squared_distance.mexa64

Great jobs!
your link with point_mesh_squared_distance.mexa64 not found(404)。could you provide a new link with the precompiled version of point_mesh_squared_distance.mexa64? Thank you very much!

And, I compile compile_gptoolbox_mex_modified will occur

>> compile_gptoolbox_mex_modified
警告: This is **VERY** experimental. In principle, this should compile all of the mex functions in this
directory. In practice, users will surely have to adjust paths and flags at the top of this file. 
> In compile_gptoolbox_mex_modified (line 96) 
Hit any key to continue...
警告: 函数或变量 'EIGEN_INC' 无法识别。 
> In compile_gptoolbox_mex_modified (line 133) 

how I solve it?

Is the predicting parameters of each primitive is independent from the other primitives

Dears,

Kindly, i want to inquire about
Is the predicting parameters of each primitive is independent from the other primitives,which means is the parameters of each primitive is independent of each other or not.how all these primitives are combined together to determine the full shape.

why the model predict primitives and at final all these primitives combines the shape,what prevent the model to predict the same primitive each time ?

why the primitives is attached to each other,what force this behavior in the prediction process

Unrecognized function or variable 'polygon2voxel_double'.

Thanks for your work!
I'm trying to run Matlab preprocessing script but have this issue reporting unrecognized function 'polygon2voxel_double'. Does anyone have solutions for this one?

Unrecognized function or variable 'polygon2voxel_double'.

Error in polygon2voxel (line 171)
Volume=polygon2voxel_double(FacesA,FacesB,FacesC,VerticesX,VerticesY,VerticesZ,VolumeSize,Wrap);

Error in precomputeShapeData (line 61)
        Volume=polygon2voxel(FV,gridSize,'none',false);

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.