Use of an autoencoding model for denoising synthetic images

Jérôme BUISINE 2a9d995fc6 Merge branch 'release/v0.0.4' 11 months ago
generate 138c254d1d Refactoring and scripts updates 11 months ago
modules @ 7e65b752b6 138c254d1d Refactoring and scripts updates 11 months ago
.gitignore b5efdd2af6 Update of ignore files from project 11 months ago
.gitmodules 138c254d1d Refactoring and scripts updates 11 months ago
LICENSE bcc604e079 First model commit 11 months ago
README.md 138c254d1d Refactoring and scripts updates 11 months ago
__init__.py 138c254d1d Refactoring and scripts updates 11 months ago
custom_config.py 138c254d1d Refactoring and scripts updates 11 months ago
dataset 19b3967aa2 Update of model architecture 11 months ago
image_denoising.py 138c254d1d Refactoring and scripts updates 11 months ago
requirements.txt 3fcd2818bf Update of dataset generation 11 months ago

README.md

Denoising with autoencoder

Description

Utilisation d'un autoencoder pour apprendre statistiquement comment il est possible de générer une image de synthèse.

Input :

  • Noisy image
  • Z-buffer
  • Normal card

or other information...

Output :

  • Reference image

Requirements

git clone --recursive https://github.com/prise-3d/Thesis-Denoising-autoencoder.git XXXXX
pip install -r requirements.txt

How to use ?

Autoencoder keras documentation

Generate reconstructed data from specific method of reconstruction (run only once time or clean data folder before):

python generate/generate_reconstructed_data.py -h

Generate custom dataset from one reconstructed method or multiples (implemented later)

python generate/generate_dataset.py -h

Reconstruction parameter (--params)

List of expected parameter by reconstruction method:

  • svd_reconstruction: Singular Values Decomposition
    • Param definition: interval data used for reconstruction (begin, end)
    • Example: "100, 200"
  • ipca_reconstruction: Iterative Principal Component Analysis
    • Param definition: number of components used for compression and batch size
    • Example: "30, 35"
  • fast_ica_reconstruction: Fast Iterative Component Analysis
    • Param definition: number of components used for compression
    • Example: "50"
  • static Use static file to manage (such as z-buffer, normals card...)
    • Param definition: Name of image of scene need to be in {sceneName}/static/xxxx.png
    • Example: "img.png"

Example:

python generate/generate_dataset.py --output data/output_data_filename --metrics "svd_reconstruction, ipca_reconstruction, fast_ica_reconstruction" --renderer "maxwell" --scenes "A, D, G, H" --params "100, 200 :: 50, 10 :: 50" --nb_zones 10 --random 1 --only_noisy 1

Then, run the model:

python image_denoising --data data/my_dataset --output output_model_name

License

The MIT license