Satellite Imagery Analysis

Overview

  1. Can use multiple types of imagery and videos :
    1. Monochromatic
    2. Multi-Spectral
    3. Color
    4. UAV Video
  2. Uses AI based object recognition to recognise multiple objects in different classes as required
  3. Works in real-time with multiple different sources
  4. Ability to integrate with 3rd party systems like C2 systems

Features

  1. Able to detect multiple target objects

  2.  On ground : Tanks, Oil Tankers, APCs, troop

  3. On water : Different types of Naval Vessels

  4. Other assets like airfields, aircraft etc

  5. Able to enhance satellite imagery by using super-resolution, through a single image or videos. Effectively decreasing Ground Sampling Distance

How it works

  1. Military Object Detection

    • Training Data obtained from a variety of different sources

    • DNN fine-tuned on the new training data and trained for detection

  2. Sat Image Super-resolution

    • High resolution training data down-sampled to mimic low-resolution images

    • DNN trained to up-sample the images using high-res images as ground
      truth