High performance computer vision systems
for real-world applications.
Portable supercomputer and 3D computer vision system
AMD Threadripper Pro 32-core CPU, ECC RAM, Nvidia RTX 3090 GPU, Nvidia A100 GPU "backpack" running Red Hat Enterprise Linux and the Nvidia Riva A.I. suite, Intel SSD running Windows, Seagate Exos 2X14 dual actuator hard drives (x8) in a 100 TB RAID 0 (> 3 GB/sec), Supermicro 4U industrial grade tower chassis with dual Platinum 1200 watt redundant PSU's, Samsung 65" 8K television (60 Hz refresh) with cart/stand, ASUS professional workstation motherboard, 100G ethernet host adapter, 10G ethernet onboard, Pelican Super-V 5U rolling rack case, dual 60 Hz industrial cameras, FLIR science kit and/or optional hyperspectral camera.
Lead time: 60 days
A system for battlefield Identification Friend or Foe (IFF) using encrypted QR codes and real-time stereoscopic computer vision instead of transponders. Relies on long-range perception, filters, and error correction when reading symbology on panels. System is supported by research in human perception concerning the recognition of faces/people at a distance. Special thanks to Dr. Geoffrey Loftus at the University of Washington for his many papers in this area. Has a wide range of defense applications. In development (expected 2024).
An adjustable cavitating hydrofoil using a combination of sensors and computer vision to decelerate ships and modify their audio signatures. In development (expected mid-2024).
An interesting SBIR solicitation that is being pursued here.
An innovative approach to using computer vision to assess mental processes and validate the system against the MMPI-3.
Computer vision as an alternative to performance motion capture of live performances.