Image Recognition

The capabilities of machine learning models have advanced substantially in recent years, making them suitable for industrial use.
There are several types of use cases where AI can be applied:

  • Image classification where a model learns to recognize an image as belonging to a certain class.
  • Object detection an object is recognized and its bounds are shown.
  • Image segmentation the shape of the object is marked and can be distinguished from its background.
  • Human pose estimation the position of the limbs, head and torso is interpreted from an image.
    This could be particularly useful for sports analytics.
  • Optical Character Recognition handwritten or typed text is recognised for digital use.
  • Anomaly detection an image is identified as being undesirably different from the requirements.

Several of these use cases can also be applied to video footage obtained from a drone.
Drones allow companies to quickly inspect vast amounts of land or infrastructure.
Manually reviewing this footage however is time consuming and tedious.
Using machine learning we can detect anomalies, objects and patterns in videos automatically.
This can be done in real time or on a stored collection of images or videos.

The following features characterize an ML2Grow solution:

  • A simple JSON service is provided or can be integrated in existing software.
  • Can handle large amounts of images.
  • Self-learning: new information is continuously used to refine the system.