Vision AI

VisionAI Multi Layered Neural Network Artificial Intelligence (AI) Imaging Toolkit
VisionAI makes machine learning of images easy. After training a VisionAI model it can be used to automatically recognize the object in images. The training data/input to VisionAI are images of any size and color depth. Each image is labeled/classified automatically. After training the AI, or with other words learning the training data/images, VisionAI can be used to make very fast predictions/image recognition (less than a second).

VisionAI also contains the following modules:

  • Extensive Image Enhancement Toolkit for improving image quality and for cleaning images with text (e.g. automatically removing lines).
  • Create an Image Collage (tiled images). Such an image collage can be used for training VisionAI.
  • VisionAI Text Recognition engine which can be used to recognize and extract text from images in several languages.
  • VisionAI Text To Speech engine which can be used to convert text to speech/audio in several languages.
  • Import images from video (local camera, internet camera, computer screen, video file).
  • Import images from a scanner.

VisionAI Artificial Intelligence Software Toolkit for Machine Learning of Images


The size of the images have an important effect on the necessary computer memory and CPU utilization, therefore VisionAI is hardware accelerated (speed up) in two ways:
  1. VisionAI takes advantage of special built-in processor (CPU) extensions. At the moment the following CPU's are supported for this optimization: all x86/x64 family processors.
  2. VisionAI supports parallel processing in case there are multiple processors and/or multiple processor cors in the computer.
 AI-TOOLKIT Download

AI-TOOLKIT Helpdesk

 Software Feedback Form

VisionAI Classify Server Included!

The VisionAI Classify server can be used to automatically process images e.g. from an other software package. You can choose the time frequency with which the server will check for new images after it is started. The other software package can then read in the results after they appear in the 'server/out' directory. In this way you can also stream new images (any amount = big data ready) to VisionAI for classification processing.

Despite the fact that neural networks modeling is a complex task VisionAI makes this task relatively simple by intuitively and automatically applying design patterns to your problem. Many of the input parameters are filled in automatically for you. One of the reasons to make specialized software packages (as DecisionAIPro, DeepAI and VisionAI) is to reduce the complexity of machine learning and to make the software easy to use and to understand.

Technical description

VisionAI is a Multi Layered Neural Network Artificial Intelligence (AI) Imaging Toolkit. VisionAI supports complex multi-layered convolutional neural networks for machine learning of images. In such an AI architecture the connectivity between the neurons is inspired by the animal visual cortex. Individual neurons respond to stimuli in a restricted region as a receptive field. The receptive fields of the neurons partially overlap and they form the visual field. The response of a neuron to stimuli within its receptive field is approximated by a mathematical convolution operation.

You can define as many neural network layers and as many filters per layer as you need. Each layer can have a different activation function which transforms/activates the pixels of the images/input data. There are three available neural network layer types:
  1. Convolution Layer - The convolution layer sweeps a 2-D filter over a batch of images, applying the filter to each window of each image of the appropriate size. You can choose the number of filters, the filter size and also the activation function of the layer.
  2. Max Pooling Layer - This type of layer makes a non-linear down sampling of the data and progressively reduces the computation in the network. You can choose the number of filters, the filter size and also the activation function of the layer.
  3. Fully Connected Layer - This is the same type of layer as in normal neural networks and it activates (with the activation function) and usually down samples the data.
Each layer’s neurons/nodes are automatically connected to the next layer's neurons/nodes and like this the neural network is formed.

 AI-TOOLKIT Download

 AI-TOOLKIT Training Video's, Decision AI Professional, DeepAI, VisionAI

VisionAI Video

VisionAIVideo can be used for detecting movement (motion) or objects in live or recorded video. Objects are detected with the so called cascade models which can be trained by VisionAI Cascade Studio. You can detect rigid or near rigid objects e.g. faces, human body, cars, animals, plants, flowers, equipment, products on production lines, etc. in video.

Visit the VisionAIVideo page for more info!

VisionAIVideo for detecting movement and objects in video


VisionAI and the VisionAI Artificial Intelligence Engine is (C) Copyright 2016-2017 Zoltan Somogyi, All Rights Reserved 


Contact

Contact Person

Address

Antwerp, Belgium

You may use LinkedIn (phone number and email available) or the contact form to the right.

Helpdesk

You may contact the AI-TOOLKIT helpdesk:


Search This Website