The journey of an AI project is an iterative one. Developers use high-level languages like Python to quickly prototype and test models, but need to convert to C code for deployment to the real world. Often, this deployment is also a part of the iterative feedback loop. AI Transformer makes this process a breeze.
AI Transformer is a cloud-based code generator for Deep Neural Network (DNN) models. Given a DNN model, it generates C code optimized for embedded systems and edge compute.
AI Transformer supports Keras(™) models. TensorFlow(™) and Theano(™) users will first need to convert their models to json and hdf5 using Keras(™).
For ANNs, it covers all the configurations for a fully connected dense layer. For CNNs, it covers standard Conv2D layer, maxpooling2D layer and flatten layer, it also covers configurations like strides, kernel_size and paddings.
AI Transformer will try to ignore unsupported features, but may result in error. If you are affected by this, please contact us for enhancing our tools.
We would love to help you out. Please contact us with as must detail as you can, and we shall get back to as soon as possible.
We will definitely love to hear from you. Contact us and we will find a way to have a conversation.