Intersection over Union for object detection





The Jaccard index, also known as Intersection over Union and the Jaccard similarity coefficient (originally coined coefficient de communauté by Paul Jaccard), is a statistic used for comparing the similarity and diversity of sample sets. The Jaccard coefficient measures similarity between finite sample sets, and is defined as the size of the intersection divided by the size of the union of the sample sets:

(source: Wikipedia https://en.wikipedia.org/wiki/Jaccard_index)

Figure 1: An example of detecting a stop sign in an image. The predicted bounding box is drawn in red while the ground-truth bounding box is drawn in green. Our goal is to compute the Intersection of Union between these bounding box.

Figure 2: Computing the Intersection of Union is as simple as dividing the area of overlap between the bounding boxes by the area of union (thank you to the excellent Pittsburg HW4 assignment for the inspiration for this figure).

Below a clear blog post about it:

https://www.pyimagesearch.com/2016/11/07/intersection-over-union-iou-for-object-detection/
I am using Intersection Over Union for testing solution to this kind of pcomputer vision problem for instance
https://www.kaggle.com/vanausloos/keras-linknet-with-intersection-over-union/edit?unified=1

Comments