grancv

Computer vision web application with granulometry purpouses, built on top of ipyvuetify, bqplot, opencv and ipycanvas.

Install

Create new conda environment:

conda create -n cv_env

Activate environment:

source activate cv_env

Install some packages:

python -m pip install -r requirements.txt

Run the app:

voila --VoilaConfiguration.extension_language_mapping='{".py": "python"}' $1 --debug --enable_nbextensions=True

Or, if you prefer the dark theme:

voila --VoilaConfiguration.extension_language_mapping='{".py": "python"}' $1 --theme=dark --debug --enable_nbextensions=True

You may want to paste this into your .bash_aliases file:

v() {
  voila --VoilaConfiguration.extension_language_mapping='{".py": "python"}' $1 --theme=dark --debug
}

Now, you can run the application with:

v index.py

Operations Available

You can use and tune combinations of the following operations:

• Neural Style Transfer

  • Select one of the available torch .t7 models available in the 'models' folder.

  • You can adjust the quality of the final result by changing the image width.

• Preprocessing

  • Filtering

    • Simple

    • Bilateral Filtering

  • Blur

    • Simple

    • Median Blur

    • Gaussian Blur

• Morphologycal Operations

  • Erosion

  • Dilatation

  • Opening

  • Closing

  • Gradient

  • Top Hat

  • Black Hat

• Thresholding

  • Simple

    • Binary

    • Binary Inverse

    • To-Zero

    • To-Zero Inverse

  • Adaptive

    • Adaptive mean c

    • Adaptive Gaussian c

To do

This application is is very initial stage. More operations will be added to the application gradually. Here are some I've thought about:

• Possibility of creating your own operation, from the combinations of wherever you want.

• Choose (and plot) the kernel. All kernel used in the operations are matrices of kernels in the form:

  [[1,1,1],
  [1,1,1],
  [1,1,1]]