Deep learning based tool for image processing. No need for Programing and GPU.
MIT License
This is an open source project to help people who are trying to use Deep Neural Network model for image processing but troubled by programming or computation resources. With DeepDIY, you can:
DeepDIY will split your training data (image + anotation file) into train set and validation set. And then pack all of data (train set, validation set and config file) into a zip file name "dataset.zip"
Download win64 portable version: https://github.com/deepdiy/deepdiy/releases
Unzip and go to
./path_of_downloaded_package/deepdiy/DeepDIY.exe
Double click DeepDIY.exe , Done!
Clone this repository
Run setup from the repository root directory
python3 setup.py install
Clone this repository
Install dependencies
pip install -r requirements.txt
Install kivy.garden.matplotlib
garden install --kivy matplotlib
For OS X users, you may need to install kivy and kivy-garden manually. The 'garden' command is available only after kivy-garden is installed successfully. Please refer to following page:
https://kivy.org/doc/stable/installation/installation-osx.html
Most of mainstream deep learning network will be included in DeepDIY, including but not limited to:
DeepDIY encourage users to share new models and weights trained on their own data.
You can train network on your own data without GPU, the training task in performed on Google Colab[https://colab.research.google.com/]. Free, fast, private and safe. Only one click, leave and drink a cup of coffee, done!
Kivy[https://kivy.org/] is a coss-platform Python framework for GUI development. Very easy to understand and use.
The core of DeepDIY software is a resource tree, which is stored in a Python dictionary. For any plugin, DeepDIY will sync resource tree dict with plugin as a property(called 'data') of plugin class. There are two type of plugins: Processing and Display.
Plugin can acquire data from tree, after running your functions, just insert result data as a child of selected node. Result can be displayed by Display plugins.
When user click a node in resouce tree, certain display plugin will be activated. Plugin can acquire selected data from the tree, and return a Kivy widget, the widget will be embedded in the window.
Peng Nie (Johnson)
Python 3.7.1, Kivy 1.10.1, TensorFlow 1.13, Keras 2.2.4, OpenCV 4.0.0.21, Numpy 1.15.4, Scipy 1.1.0, Matplotlib 3.0.2
MIT is licensed under MIT license.