Deep fully-connected neural network, Deep Convolutional network, and Deep Recurrent network implementation using OpenCV & OpenCL backend.
The nnet framework was written with support for NIFTI medical imaging formats, to enable deep-learning applications development with 3D volumetric medical imaging data (MRI, CT, PET, SPECT, DWI). Provided primatives enable users to build custom CNN/RNN/FC models for classification, segmentation, and feature detection.
Project inspired by homeworks in Andrew Ng's Deep Learning specialization through deeplearning.ai and Coursera