Image classification
Image Classification is a computer vision technique to label an image from a set of predefined classes.
Object detection is a computer vision teqhnique for locating objects in an image.
For us is very easy to say what is in a picture, but for a computer, it is very challenging. In image classification one aims to assign a label to an image from a set of predefined classes. Ex. say one wants to classify what dog species it is in a picture, one decides on what dog species one wants to be able to detect, then the model will learn to pick the most likely classification.
Image classification is well suited in situations where one needs to assign a single class to the whole image and one does not need to locate where in the image something is. For example, say one has medical images off chest x-rays one can use it to say if the image shows cancer yes or no.
There are also extensions of image classification to assign multiple classes to the same image which is referred to as multiclass classification. If one actually needs to locate objects in an image something like Object Detection, Semantic Segmentation or Instance Segmentation is a better technology to use.
To get a better understanding of how uncertainty can add value to image classification, have a look at our blog post on