Though subtle, there is a difference between computer vision and image recognition. Therefore, we decided to explain the nuance between these two often associated and confused concepts.
Computer vision and image recognition are not the same things!
As explained in a previous article, computer vision is a branch of artificial intelligence (AI). Indeed, it is the eyes of AI. More specifically, computer vision is a set of techniques allowing the automation of tasks from an image or video stream.
Image recognition is a subset of computer vision. It consists of a set of techniques for detecting, analyzing, and interpreting images to favor decision-making. It works through a neural network trained via an annotated dataset.
Here is a diagram to help you understand the hierarchy between these different fields of study.
The purpose of image recognition is similar to that of computer vision, i.e. to automate the performance of a task. In image recognition, these tasks are varied. For instance, they can be the labeling of an image through tagging, the location of the main object of an image, or guiding an autonomous car. We then talk about image classification, object detection, segmentation or tagging.
Computer vision is not just image recognition!
As you can see from the diagram above, computer vision is not only about image recognition. Indeed, computer vision also encompasses optical character recognition (OCR), facial recognition and iris recognition.
OCR, or text recognition, allows the translation of printed, typed or handwritten texts into computer text files.
On the other hand, facial recognition consists of the automatic recognition of a face within an image to determine its identity. The main applications are in video surveillance, biometrics, and robotics.
Similarly, iris recognition is a biometric technique that also allows identifying a person through the iris. Indeed, the iris, the colored part of the eye, is composed of many complex patterns that make it different and unique to every person.
Image recognition in practice
To illustrate our explanations and help you better understand the idea of image recognition, here is an example of a practical case study we worked on at Deepomatic.
The safety of users in their car parks is a key concern at Indigo. This is why Deepomatic has developed an intelligent video surveillance system based on image recognition. This device is able to detect threats, camera occlusions or emergency situations. Indeed, the system can recognize hooded people, people holding a suspicious object, people in distress, etc.
If you are interested in the subject, please refer to our white paper From video surveillance to smart cameras.
The possibilities of image recognition are not limited to this. In fact, there is a wide range of other applications in various industries such as automotive, construction, pharmaceutical, etc.
To follow the news, challenges, and opportunities surrounding image recognition, do not hesitate to consult the blog of Augustin Marty, Deepomatic’s CEO, on L’Usine Nouvelle.
