Study of synthesis images noise detection using 26 attributes

Jérôme BUISINE 249da2dc15 Merge branch 'release/v0.0.8' 5 anos atrás
analysis cb6026f2c7 Add of 26 features metric 5 anos atrás
data_processing cf7e7e11b5 Update of generate data scripts 5 anos atrás
display 49282f8488 Update of folders architecture 5 anos atrás
generate cf7e7e11b5 Update of generate data scripts 5 anos atrás
modules @ d5de038bdc ac0626b314 Add of modules and others scripts 5 anos atrás
others cf7e7e11b5 Update of generate data scripts 5 anos atrás
prediction 49282f8488 Update of folders architecture 5 anos atrás
run cf7e7e11b5 Update of generate data scripts 5 anos atrás
simulation cf7e7e11b5 Update of generate data scripts 5 anos atrás
.gitignore cf7e7e11b5 Update of generate data scripts 5 anos atrás
.gitmodules ac0626b314 Add of modules and others scripts 5 anos atrás
LICENSE dc0463b6b5 Project initialization 5 anos atrás
README.md ac0626b314 Add of modules and others scripts 5 anos atrás
custom_config.py 49282f8488 Update of folders architecture 5 anos atrás
data_attributes.py ac0626b314 Add of modules and others scripts 5 anos atrás
models.py 49282f8488 Update of folders architecture 5 anos atrás
requirements.txt 4093aaaf5c Dependencies and documentation updates 5 anos atrás
train_model.py 4093aaaf5c Dependencies and documentation updates 5 anos atrás

README.md

Noise detection using 26 attributes

Description

Noise detection on synthesis images with 26 attributes obtained using few filters.

Filters list:

  • average
  • wiener
  • median
  • gaussian
  • wavelet

Requirements

pip install -r requirements.txt

Generate all needed data for each features (which requires the the whole dataset. In order to get it, you need to contact us).

python generate/generate_all_data.py --feature all

How to use

Multiple directories and scripts are available:

  • dataset/*: all scene files information (zones of each scene, SVD descriptor files information and so on...).
  • train_model.py: script which is used to run specific model available.
  • data/*: folder which will contain all .train & .test files in order to train model.
  • saved_models/*.joblib: all scikit learn models saved.
  • models_info/*: all markdown files generated to get quick information about model performance and prediction. This folder contains also model_comparisons.csv obtained after running runAll_maxwell.sh script.
  • modules/*: contains all modules usefull for the whole project (such as configuration variables)

Remark: Note here that all python script have --help command.

python generate_data_model.py --help

python generate_data_model.py --output xxxx --interval 0,20  --kind svdne --scenes "A, B, D" --zones "0, 1, 2" --percent 0.7 --sep: --rowindex 1 --custom custom_min_max_filename

Parameters explained:

  • output: filename of data (which will be split into two parts, .train and .test relative to your choices).
  • interval: the interval of data you want to use from SVD vector.
  • kind: kind of data ['svd', 'svdn', 'svdne']; not normalize, normalize vector only and normalize together.
  • scenes: scenes choice for training dataset.
  • zones: zones to take for training dataset.
  • percent: percent of data amount of zone to take (choose randomly) of zone
  • custom: specify if you want your data normalized using interval and not the whole singular values vector. If it is, the value of this parameter is the output filename which will store the min and max value found. This file will be usefull later to make prediction with model (optional parameter).

Train model

This is an example of how to train a model

python train_model.py --data 'data/xxxx' --output 'model_file_to_save' --choice 'model_choice'

Expected values for the choice parameter are ['svm_model', 'ensemble_model', 'ensemble_model_v2'].

Predict image using model

Now we have a model trained, we can use it with an image as input:

python prediction/predict_noisy_image_svd.py --image path/to/image.png --interval "x,x" --model saved_models/xxxxxx.joblib --feature 'lab' --mode 'svdn' --custom 'min_max_filename'
  • feature: feature choice need to be one of the listed above.
  • custom: specify filename with custom min and max from your data interval. This file was generated using custom parameter of one of the generate_data_model*.py script (optional parameter).

The model will return only 0 or 1:

  • 1 means noisy image is detected.
  • 0 means image seem to be not noisy.

All SVD features developed need:

  • Name added into feature_choices_labels global array variable of modules/utils/config.py file.
  • A specification of how you compute the feature into get_svd_data method of modules/utils/data_type.py file.

Predict scene using model

Now we have a model trained, we can use it with an image as input:

python prediction_scene.py --data path/to/xxxx.csv --model saved_model/xxxx.joblib --output xxxxx --scene xxxx

Remark: scene parameter expected need to be the correct name of the Scene.

Visualize data

All scripts with names display/display_*.py are used to display data information or results.

Just use --help option to get more information.

Simulate model on scene

All scripts named prediction/predict_seuil_expe*.py are used to simulate model prediction during rendering process. Do not forget the custom parameter filename if necessary.

Once you have simulation done. Checkout your threshold_map/%MODEL_NAME%/simulation_curves_zones_*/ folder and use it with help of display_simulation_curves.py script.

Others...

All others bash scripts are used to combine and run multiple model combinations...

License

The MIT license