Object detection

What is object detection?

Object detection, a subcategory of computer vision (CV) and artificial intelligence (AI), is the ability for computer technology to identify and locate objects in an image or video through the use of algorithms and machine learning. The goal of object detection is to replicate the human ability to look at images or video and recognize and locate objects of interest quickly.

Object detection is often mistaken for image recognition, however, these two terms have different meanings. Image recognition simply assigns a label to an image, whereas in object detection, the model is able to identify where each object is in the image and what label should be applied. For example, an image of a cat can be labeled “cat” using image recognition, but an image of two cats will also be labeled “cat.” Object detection, by comparison, is able to provide more information about these images.

Object detection can use either of the following methods:

• Machine learning-based approach: Features (color, edges, corners, etc.) are manually extracted from various images and added to a machine learning model. This data is used to train the model to predict what category test images would fit into.
• Deep learning based-approach: Training images are fed directly into a convolutional neural network (CNN) — the most common deep learning architecture — where the feature learning and classification task is automatically completed.

Benefits of object detection

Object detection technology can be found in applications such as autonomous vehicles, video surveillance, anomaly detection and more. Some of the benefits of this technology include:

• Efficiency: Business processes can be automated using object detection systems. For example, in the medical field, object detection models can be trained to automatically identify skin lesions, which can help doctors to more quickly and accurately make a diagnosis.
• Information availability: Object detection can provide organizations with information that they may not have previously been privy to. This information can be leveraged to help with decision-making. For example, in agriculture, an object detection model can help identify and locate potential instances of disease, allowing farmers to quickly and more accurately detect threats to their crop and take the appropriate action.