Visual Genomes Foundation

Join the Exascale Visual AI initiative to sequence Visual DNA from images using Volume Learning to enable new applications!

Visual AI + visual computing

Technology Expertise

Visual AI,  computer vision, imaging and cameras, robotics & automation, depth sensing, forensic and expert image analysis, media cryptography

Engineering

Systems design, architecture, HW/ SOC/ FPGA, software development, embedded systems, apps, libraries, optimizations for accuracy, power, performance

Business Development & IP

Competitive analysis, market research, IP portfolio analysis, patent generation and IP development

About

Experience

Krig Research is a pioneer in visual AI, computer vision, machine learning, advanced imaging, and robotics systems, providing complete solutions for clients ranging across a wide range of governments and industries, world-wide. Small and large projects are welcome.

Confidence

An ongoing business relationship and continued business are mutually beneficial, so every effort is made to understand your requirements and provide the best possible solution within your project parameters. Find out if Krig Research has the expertise and experience you need.


Trust

A free confidential consultation is available on request. Krig Research provides complete engineering services to complete any  visual computing project. In addition, specific projects can be executed to enhance your business via competitive market analysis, patent portfolio development, product positioning, technology briefings, and special reports.



Products

NeuroMatrix(tm)

The NeuroMatrix(tm) system contains a working model of the visual cortex of the human brain, including models of the eyes, early visual processing centers, and higher level reasoning centers. Capable of learning precise signatures of individual objects.

NeuroTarget(tm)

The NeuroTarget(tm) system locates targets in the visible  light spectrum, therefore is capable of surviving common electronic warfare countermeasures which affect radar-based systems. 

VIsual Genomes Foundation

The massive project to learn visual DNA and visual genomes from hi-res images, and provide free visual AI products and services to the public. Sponsors and members are encouraged to apply - learn more below.

Resources

Computer Vision Metrics

One of the most complete resource texts on computer vision. Download a complimentary copy of the 500 page classic from the Embedded Vision Alliance or APRESS.

Textbook Edition

The classic 653-page textbook edition in English from Springer-Verlag has also been translated into Chinese, updated with a comprehensive survey of deep learning, DNNs, RNNs, and historical computer vision background.

Synthetic Vision using Volume Learning and Visual DNA

Ground-breaking text describing synthetic vision using Volume Learning (not deep learning) and Visual DNA, selected by the Visual Genomes Foundation, from DeGruyter Press.

OpenVDNA - the visual genomes foundation

Visual DNA - OpenVDNA Catalog - OpenVDNA Search - OpenVDNA Toolkit

The OpenVDNA initiative is the work of the Visual Genomes Foundation, providing public resources for visual AI solutions. Visual DNA are taken from thousands of small pieces of images, like puzzle pieces composing the fabric of the image. Visual DNA compose a rich feature space represented by over 16,000 individual metrics in four primary bases: Color, Shape, Texture and Glyphs. The VGF provides OpenVDNA resources:


  • The OpenVDNA Catalog is a public collection of Visual DNA sequenced from hi-res 4k RGB images stored as petabytes of Visual DNA features, including Learning Agents that define strands and bundles of related Visual DNA, inspired by the human DNA catalogs produced by the Human Genome Project.
  • The OpenVDNA search engine allows the public to submit images for analysis against the Visual DNA catalog, to discover the Visual DNA in images, and find images with similar Visual DNA. However, VGF sponsors and members can create application-specific visual DNA catalogs for commercial and research purposes.
  • The OpenVDNA Toolkit enables application developers and researchers to access the OpenVDNA catalog running on a supercomputer cloud, and connect remote endpoint devices such as drones, smart phones, and IoT devices.

Synthetic Vision and Volume Learning

A Synthetic Vision model is used to perform Volume Learning, to gather a large volume of multivariate and multidimensional visual DNA features from each image. Synthetic Vision is described in the De Gruyter book "Synthetic Vision Using Volume Learning And Visual DNA", informed by neuroscience models, deep learning models, and computer vision research. While deep learning models exist in a 1D array of feature weights, volume learning produces a multivariate and multidimensional features space supporting nearly 26,000 different feature metrics for each visual DNA feature.


Synthetic Vision and Visual Cortex Memory

The synthetic vision model includes a photographic visual memory for each visual cortex feature region V1..Vn. A neural feature space is organized using content addressable memory (CAM), and neural clusters, following neuroscience research such as the Hubel and Wiesel model of neural connectivity. Each visual DNA feature exists in a visual cortex model that includes feature-specific correspondence metrics.

Learn more

Contact

Let's discuss your project.