What is Image Segmentation?

Back

Oct. 20, 2021

You might know you want to bring computer vision into your organization—but do you know exactly which computer vision use case would offer the most benefit? Image segmentation is a subfield of computer vision that seeks to divide an image into contiguous parts by associating each pixel with a certain category, such as the background or a foreground object.

The output of image segmentation is a “masked” image in which each pixel is given a certain color based on the object it is a part of. For example, in a photograph of a street scene, one person may be colored blue, another person may be colored green, a car may be colored red, the background may be colored white, etc.

Image segmentation is related to, but distinct from, other computer vision tasks such as image recognition and object recognition. To illustrate the difference, suppose that we have an input image of a soccer game, with multiple players on the field. The results of running these three tasks on this image would be as follows:

  • Image recognition: A short description of the entire image—“soccer game.”
  • Object recognition: Bounding boxes around each object in the image—the players, the soccer ball, the goalposts, etc.
  • Image segmentation: An output image in which all the objects (and the background) have been assigned a different color, creating a flat, cartoonish appearance.

The potential applications of image segmentation include:

  • Healthcare AI: Medical diagnostic imaging (CT scans, X-rays, etc.) can leverage image segmentation in order to separate different areas of concern, such as cancerous vs. non-cancerous cells or tumors.
  • Geospatial AI: Image segmentation can be applied to geospatial images, such as those from drones and satellites, to identify different geographical features or regions.
  • Self-driving cars: One of the largest possible use cases for image segmentation is autonomous vehicles. With input from a camera mounted on the vehicle, AI models perform real-time image segmentation to remove clutter and identify the most salient parts of the image (e.g. pedestrians, other cars, road signs, etc.).

Our computer vision glossary contains all the computer vision definitions you need to get up to speed. Thinking about event detection use cases for your own business? We can help. Get in touch with our team of AI experts for a chat about your situation, or sign up today with a free account on the Chooch computer vision platform.

×

Oops, something went wrong.
Please try back later.

Click here Tap here to upload Upload your image
and Chooch will recognize it.
Supported formats are .jpg .jpeg .png

×