Overview


Growing Image and Audio based AI applications with neural network (NN) and machine learning algorithm processing at the edge devices has increased need for AI Computing in low power embedded devices. Customers need to build their own custom AI Training models so that quality AI inference models are generated to be used by these AI edge devices.
Using our AI Work Flow, customers can build their suitable custom AI Training models for advanced AI/ML based Image and Audio classification applications.

Customer Challenges


At present, there are few software tools that can obtain pre-trained neural networks from cloud. But building application specific AI Training models for cloud training to realize AI computing will help the customers to build their own AI inference models for their specific applications.

Using our AI Work Flow, customers can build their suitable custom AI Training models for advanced AI/ML based Image and Audio classification applications. These models can be trained on cloud for AI algorithms for a series of Neural Network (NN) frameworks such as Caffe 2, TensorFlow Lite and Arm NN, to generate inference engines. 

The AI Work Flow can avoid pre-training neural network models borrowed from the cloud. The developers can create data models based on Images and Audio signals. 

Eoxys AI Work Flow Tool


Using our AI Work Flow, developers can build suitable custom AI Training models for advanced AI/ML based Image and Audio classification applications. The developers can create data models based on Images and Audio signals. Developers can build their AI Training models so that quality AI inference models are generated to be used by these AI edge devices.


Data preparation.

This is the sample GUI screen of Image data preparation. And there is several steps like uploading, Image-pre-processing, Augmentation and finally we will get the processed sample data.




Tile processing component.

This is the sample GUI screen of Tile processing component. And This block will useful to get number of tiled input images on rows and columns. This is one of the Pre-processor blocks. We are giving number of rows and number of columns as input to this block aling with the input images.




Grey scale processing component.

This is the sample GUI screen of GREY scale processing component. And This block will process the colored sample Images to Greyscale Images. This is Augmentation block. We are giving Percentage of input images to be converted to greyscale as a input to this block along with the input images.




Blur processing component.

This is the sample GUI screen of BLUR processing component. And This block will useful to get the required amount of blurrness for input images.This is a Augmentation block. We are giving kernel size and percentage of input to be get blurred as input to this block along with the input images.




Brightness processing component.

This is the sample GUI screen of BRIGHTNESS processing component. And This block will useful to get the required amount of brightness for input images. This is a Augmentation block. We are giving percentage of brightness and percentageof input to be get brighter as inputs to this block along with the input images from source block.




Hue processing component.

This is the sample GUI screen of HUE processing component. And This block will useful to change the hue of input images. This is a Augmentation block. We are giving Percentage of hue to be applied for the images and percentage of inputs to be applied as input to this block along with the input images.