Computer Vision Simplified
A simplified explanation of computer vision and its use in regards to security and facial recognition
From being used within iPhones to roadside cameras, the number of applications of computer vision has skyrocketed over the past decade. Computer vision, often abbreviated CV, is a technology which assists computers with visualizing and understanding the world around us. Due to widespread use of new mobile technology, affordable and accessible computer power, and new computer algorithms such as convolutional neural networks that can take advantage of the latest, superior hardware, a new technological renaissance has commenced within the computer vision field. Today I will be explaining to you the intricacies of this emerging field in a more conventional and simplified manner.
Generally, computer vision can be categorized as a subset of machine learning and AI, which uses computer algorithms to function. The objective of computer vision is to understand the content of digital images and videos. By replicating human vision and subsequently analyzing the captured content, the process of analyzing the content of the captures is done by interpreting or detecting a description from the image or video. Specifically, the long short term memory (LSTM) aspect of a recurrent neural network (RNN) is used to accomplish this. A network is setup using the LSTM architecture of an RNN. This is done to get a landmark estimation for a face. A face landmark estimation is where the key points of a face such as, dimples, blemishes, and eye color, are identified. These neural networks are “trained” in a way to prepare for practical use. By running the AI sample tests in which it is presented thousands of sample images, the AI’s algorithm begins developing and expanding its scope of knowledge. This process helps with comprehending and breaking down everything that is contained within an image such as pixels. By scanning almost every pixel of which an image is comprised of, the AI begins identifying certain patterns images and begins to memorize or remember them for future situations. As mentioned before, in regards to facial recognition dimples or birthmarks can serve as these patterns. For autonomous driving or medical scans, the AI can also be trained to carry out an optimal output action or an ideal action for the situation it is in. The optimal action helps the AI easily and safely carry out its job. For example an autonomous vehicle may beep or skew to a certain direction based on how it is to another vehicle, a person, or any other potential obstacle.
In regards to defense and security, computer vision is becoming more and more prominent. From casinos to banks and other high-security environments, image and facial recognition systems have been known to stop crimes and identify criminals. In a 2014 robbery, NEC NeoFace, Chicago’s $5.4 million facial recognization system, helped police identify and arrest 22 year old Pierre Martin. Video surveillance and NeoFace had caught Martin robbing a train passenger and getting away with the individual’s smart phone. Using this technology police were able to quickly identify and apprehend Martin. Anita Alvarez, the former State’s Attorney for Cook County, Illinois would go on to say, “This case is a great example that these high-tech tools are helping to enhance identification and lead us to defendants that might otherwise evade capture”, solidifying the role of facial recognition within the city of Chicago and in the field on security.
Helping artificial intelligence analyze the data it collects using computer vision has always been a challenge for developers and computer engineers. From not having an adequate understanding of ophthalmology in collaboration with neurology to the presence of the many complexities within the visual world, helping computers see and analyze the data collected has always been a challenge. Despite this, many of the upcoming applications of computer vision truly redefine how this field of study and innovation is perceived. From having the ability to diagnose patients through CT scans to guiding autonomous vehicles by identifying road signs, people, traffic lights, and other obstacles, a new chapter of computer vision technology is beginning to emerge.
Stay in Touch
Thanks for reading! I hope you enjoyed the article!
Connect with me on Linkedin
Follow me on Medium
Feel free to reach out by e-mail with any questions or inquiries: maharshih.barot@gmail.com
👏 if you enjoyed the article!
Many fascinating articles to come, so stay tuned!
