Skip to content

Computer vision

Technology uses computer vision technology based on convolutional neural networks. These learning algorithms, which mimic the functioning of the human brain, are designed to extract important data within images before proceeding to their analysis. This allows them to process rich visual signals with less computational power than a network that would process each pixel independently.

Computer vision has many practical applications. Some examples to get an idea:

  • Recognize cars on a video, draw a frame or segment the image to identify their position and movement over time.
  • Identify a tree species from a photo.
  • Detect the presence, or not, of markers of a pathology on medical images.
  • Detect the presence, or not, of defects on the surface of industrial parts.
  • Identify the position of an object to allow a robot to catch it.
  • Detect an event, for example the start-up of a machine without safety barriers.
  • Enable autonomous vehicles to navigate in space while taking into account their environment.
  • Reconstruct 3D surfaces from 2D images.

Among the tools available for these treatments, has acquired a good mastery of PyTorch, the Swiss Army knife of neural networks, and Yolo, which allows to perform object detection in real time on a video stream. Many other tools related to statistics and image processing are also integrated.

Leave a Reply