Sep. 26, 2021
Computer vision, and subfields of computer vision such as object detection and object recognition, can certainly be classified under machine learning, depending on how you build the computer vision model. IBM defines machine learning as“a branch of artificial intelligence (AI) and computer science which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy.”
Nearly all cutting-edge work done in object detection and other computer vision tasks these days is accomplished using machine learning. Generally, you start by creating a dataset (or using a preexisting one) that contains the objects you want the model to detect. Then, the model is trained on this dataset, learning by example by recognizing the correct answers over time. This is typically done through backpropagation, i.e. fine-tuning the weights of a deep learning model such as a showing convolutional neural network (CNN). When the model emits an incorrect answer, the model’s weights are slightly adjusted in a way that will hopefully make it more likely to output the correct answer—which is precisely how machine learning works.
The best way to think of the relationship between computer vision and machine learning is a Venn diagram, with both fields representing intersecting but non-equal sets: