Tactical Tailor

DECISIVE ANALYTICS Corporation Wins Navy SBIR PHI Computer Learning Obfuscating Adversarial Kit (CLOAK) Contract

ARLINGTON, Va., Oct. 16, 2018 /PRNewswire/ —DECISIVE ANALYTICS Corporation (DAC), a leader in cutting-edge Machine Learning development and application, has been awarded a Small Business Innovation Research (SBIR) Phase I contract by the Office of Naval Research (ONR) to develop the Computer Learning Obfuscating Adversarial Kit (CLOAK) system. Under this effort, DAC will provide the Navy with a portable camouflaging kit by advancing state-of-the-art adversarial imagery though physical-world manipulation and data synthesis.

In Phase I of this effort, DAC will select and train two modern classifiers on military and commercial vehicles, develop an advanced data synthesis pipeline, and develop an algorithm for learning camouflage configuration. DAC’s algorithms will combine advanced 3D data synthesis with a constrained adversarial generation process for generating physical-world robust computer vision camouflage.

5 Responses to “DECISIVE ANALYTICS Corporation Wins Navy SBIR PHI Computer Learning Obfuscating Adversarial Kit (CLOAK) Contract”

  1. Hubb says:

    This sounds sci-fi! Please add video or pictures in coloring book format for those of us that do not have a physics degree.

  2. Desert Lizard says:

    Even the meaning of this article is camouflaged.

  3. Joe says:

    Think we can file a FOIA on whatever this is to obtain a plain-English translation?

    I did a light search and this has been out for a few days, but nothing clarifying it is turning up.

  4. AbnMedOps says:

    Reminds me of an old MAD magazine cartoon about the “Fort Nixon National Camouflage Center”.

  5. rrossouw says:

    That is plain English…

    To re state it.
    They’re going to train AI’s to determine how the 3D shapes of objects (i.e. tanks, ships, etc.) have to be “adjusted” to either avoid detection or identification – both if they apply Lt.Col O’Neill’s 1970’s Dual-Tex tech – by human and computer observers

    The kit with which the objects will be modified is unknown.
    Consider artillery camouflage netting, consider how the simple profile changes if variable length poles are used to create surface roughness of the netting tent erected, thereby bettter approximating the environment.