BY IQRA
Computer Vision is a branch of Artificial Intelligence that allows computers to comprehend and evaluate the physical world. For decades, computer vision, or the ability of artificially intelligent systems to “see” like humans, has piqued interest and sparked extensive research. The goal of computer vision research is to create machines that can automate tasks that involve visual cognition by simulating the human visual system. The initial steps in computer vision were taken in the 1950s, when early neural networks began to recognize the edges of objects and categorize them according to their shapes. Using optical character recognition, the first commercial Computer Vision applications were used to interpret printed text for the blind in the 1970s (OCR). Large data sets of images became available online for study as the internet grew in the 1990s, propelling the development of facial recognition software, algorithms were devised to meet distinct issues across a wide range of sectors as Computer Vision applications expanded, and they have continuously become more sophisticated over time. Smartphone cameras are now ubiquitous, ensuring a steady stream of photographs and videos, and Computer Vision technology has become more widely available, making it even more tempting to businesses.
Computer vision technology of today uses Deep learning techniques that use a special type of neural network called a convolutional neural network (CNN) to make sense of images. Thousands of sample images are used to train these neural networks, which helps the algorithm analyse and break down everything in an image. These neural networks “memorise” patterns by pixel-by-pixel scanning of images. It also remembers the optimum output for each input image (in the case of supervised learning) or classifies image components based on scanning properties like contours and colours. The systems then use this memory as a reference point while scanning further images, and with each iteration, the AI system improves its ability to provide the right output.
Object recognition and classification accuracy rates have risen from 50% a decade ago to 95% in 2019, and today’s technologies are even more accurate than humans. Artificial vision technology is having an impact on a variety of fields that use computers to evaluate images. The military, industrial, healthcare, automotive, data, and retail industries are among them. As computer vision technology advances, the applications it has are numerous, few applications are discussed here.
In Healthcare: 90 percent of all medical data is image-based, there have been a slew of Computer Vision applications in the healthcare industry. The system can detect problems in MRI and CAT scan imagery with a significantly better degree of accuracy than medical experts can. Radiologists, cardiologists, and oncologists have embraced Computer Vision because it can detect early-stage cancers, arteriosclerosis, and thousands of other disorders. It’s also become an important part of a lot of invasive surgeries. Gauss Surgical, for example, has created a technology that detects the amount of blood on surgical sponges and tracks blood loss in real time. Currently, the technique is being employed in operations such as caesarian deliveries.
In military: Computer Vision is a critical enabling technology for modern armies, as it aids security systems in detecting enemy forces or traitors and improves the targeting capabilities of directed missile systems. Image sensors are significantly employed in military concepts such as situational awareness to offer combat knowledge used for tactical decision-making. Human drivers and pilots can also benefit from computer vision systems, which help them avoid hostile fire.
In Autonomous Cars: Driverless automobiles, which rely heavily on Computer Vision and Deep Learning, are one area that has captivated the public’s curiosity. While autonomous vehicle technology hasn’t quite reached the stage where it can completely replace human drivers, it has come a long way in recent years. Artificial Intelligence (AI) collects data from millions of drivers, learning from their actions to automate lane finding, predict road curvature, detect hazards, and understand traffic signs and signals. Waymo, for example, has driven seven million miles on public roads to train its Computer Vision algorithms.
The market for Computer Vision applications is rapidly expanding, and as the technology gets more inexpensive, we can expect to see an increase in the usage of image identification and Deep Learning in Computer Vision, assisting in the realization of the smart city goal.
Iqra, a Gold Medalist , is Research Scholar at Department of Computer Science,Islamic University of Science and Technology