We develop portable deep learning and artificial intelligence solutions that make exploring the model space more efficient and solutions more quickly deployable. But you've probably heard so many similar claims from different startups that its all a blur by now.

It might be worth checking out the samples below, which are trained on a Windows machine (with no need for CUDA, PyTorch or even coding) on NVidia, AMD or Intel graphs. And the trained weights are directly ported to inference apps available in the Android and iOS stores.
With the fine-grain object classification sample, users can train the bottom layers of the network to detect new classes of objects (on any Windows machine with a modern graphics card) and then distribute the trained weights as plugins to the accompanying iOS and Android apps.


Sample Projects


Composable Deep Filters and Style Transfer

As opposed to regular deep filters, which leave little room for human involvement in their design, Blurate filter descriptions incorporate functions conditioned on the presence or absence of deep features. These filter descriptions can be shared across iOS and Android devices. StylePlay allows you control how style is applied to a content image.


Fine-Grain Object Classification


With this app (for iOS and Android) you can train fine-grain classification for targeted categories and deploy the model to customer’s mobile devices.



contact: admin @ our domain



We don’t use cookies or collect personal data, primarily because you’re not that interesting.

We do use google analytics though, just because it’s amusing to see what type of folk come here.