sciTorch module is built to allow more non-programmers (light-programmers) to be
able to use, and understand deep neural network. The reason of choosing libTorch
(pyTorch) as the engine is because it provides python/C++ parity (in the
roadmap) which make it much simple to be integrated seamlessly with Scilab.
This initial release wouls start a baby step of integration, which is importing
thr torch script module from the pyTorch into Scilab, and used it for
inferencing. Working with Scilab IPCV 4.1.2, it make the image preprocessing and
post processing much simpler to prepare to the image for the network, as well as
to present to the users.
This makes the module suitable for:
1. Lecturers who want to show the use of pre-trainined DNN/CNN
2. As the option to test the network before deploying to other platform
3. Understanding the how the correct input data important (data type, data
range, normalizing data, etc) to get good result
4. Simplifying the image processing and computer vision parts with the help of
Scilab IPCV module
We wish more experts would contributes to this module and make it a more
complete high level tools for Deep Neural Network.