27
$\begingroup$

I'm fairly new at computer vision and I've read an explanation at a medium post, however it still isn't clear for me how they truly differ.

Improve this question
$\endgroup$

2 Answers 2

Reset to default
34
$\begingroup$
  • Object Detection : is the technology that is related to computer vision and image processing. Its aim? detect objects in an image.
  • Semantic Segmentation : is a technique that detects , for each pixel , the object category it belongs to. All object categories ( labels ) must be known to the model.
  • Instance Segmentation : same as Semantic Segmentation, but dives a bit deeper, it identifies , for each pixel, the object instance it belongs to. The main difference is that it differentiates between two objects with the same label.

Here's an example of the main difference.

Differences

In the second image where Semantic Segmentation is applied, the category ( chair ) is the output class, all chairs are colored the same. In the third image, the Instance Segmentation, goes a step further and separates the instances ( the chairs ) from one another in addition to identifying the category ( chair ) in the first step.

Hope this clears it up for you a bit.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you for your answer! You've clarified it for me! To make sure I understand, could I say that both type of segmentations are object detection techniques and that instance is a "higher form" of segmentation, since it does not only segment an object from others categories, but also between each instance of its own category? $\endgroup$
    – Guilherme Marques
    May 15, 2019 at 17:53
  • $\begingroup$ Exactly, thats it! $\endgroup$
    – Blenz
    May 16, 2019 at 8:21
1
$\begingroup$

Instance segmentation:

  • The combination annotation of target detection and semantic segmentation.
  • The target detection comes first, and then each pixel is labeled (semantic segmentation).

Compared to the image above, we take the person as the target objection for example:

  • Semantic segmentation does not distinguish different instances in the same category (all people are marked red)

  • Instance segmentation distinguishes different instances of the same category (different people are distinguished by different colors)

I recommend you to have a look at this article: What is Semantic Segmentation, Instance Segmentation, Panoramic segmentation.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.