Image recognition today
Thanks to the speedy progression of Artificial Intelligence (A.I.), particularly in recent years, there have been numerous applications of image recognition cropping up in various fields. Some of them you will already know about, but here are some you may not be so familiar with and some you have probably never heard of at all.
1. Marriage registration
One of the most popular applications of image recognition technology is facial recognition. In fact, in 2017, the global facial recognition market was valued at over $4 billion and is predicted to reach $10 billion by 2023. Facial recognition technology has been used by many for various use cases. But one particular example might surprise you: facial recognition has helped to carry out marriage registration.
In January 2018, authorities in Chongqing, China introduced facial recognition for marriage registration in order to make the process easier and shorter. Once a picture is taken, the facial recognition system selects a face and verifies it against the data from other official documents and the database of the public security bureau. The reason why facial recognition is more efficient than humans for this particular process is due to the speed at which it can carry out ID checks. It can complete an ID check in just 0.3 seconds, but it would typically last up to 10 minutes without such technology.
2. Behavior monitoring
Another way in which facial recognition has been used is through behavior monitoring for students in schools. In Hangzhon, China, a high school uses a facial recognition camera which scans the classroom every 30 seconds. The system is designed to record and log actions such as writing, reading, standing, hand raising, listening to the teacher, and leaning on the desk, as well as the emotions of the pupils. The system also allows students to purchase lunch in the canteen, as well as goods at the school shop and borrow books from the school library.
3. Food recognition
An even more peculiar use case for image recognition is food recognition technology. An example of this application would be the technology from the company SRI which analyzes images of food to provide nutritional information using patented computer vision technology. It has two capabilities: food recognition and volume estimation, both of which can be utilized to collect information about the food, portion size, and nutritional value.
Additionally, the technology also uses contextual information, such as location, to consolidate data from menus from the restaurants in which the image was taken. Food recognition seems to be a popular application of image recognition technology as there are a lot of examples of companies taking advantage of it, such as Calorie Mama, Foodai, SNAQ, and Bite AI.
4. There’s Waldo
If you thought those use cases were strange, this next one will surely question your perceptions of Artificial Intelligence and its limitations. Image recognition has also been used to find Waldo from ‘Where’s Waldo?’. Yes, really. The There’s Waldo robot was built by marketing agency Redpepper. It uses Google’s AutoML Vision to find and identify Waldo by feeding it images of the illustrated character. Afterward, a robotic arm containing a Vision Camera Kit photographs the pages of the puzzle book and compares it to the model using OpenCV. Once it finds a match with at least 95% confidence it points to Waldo. While this particular image recognition use case may not be the first thing you would think of in terms of utilizing the technology to achieve an important task, it nonetheless showcases what Artificial Intelligence is capable of as well as the accuracy at which it can perform those capabilities.
One of the most important jobs image recognition has been able to assist in accomplishing is diagnosing diseases. The Chinese startup Infervision uses image recognition technology to diagnose signs of lung cancer in patients using x-rays. It would usually take a medical professional 15-20 minutes to examine one image. However, Infervision’s technology can process the images and formulate a report in just under 30 seconds.
Furthermore, a recent Stanford University Research report by Sebastian Thrun found that a deep learning algorithm is now able to diagnose potentially cancerous skin tumors as accurately as a board-certified dermatologist. This is an incredible leap forward for image recognition capabilities. Some researchers even contend that a cure for cancer assisted by Artificial Intelligence is less than a decade away.
These are just a fraction of the image recognition use cases you probably haven’t heard of, but there are many more applications of A.I. at our disposal. There is a plethora of companies, businesses, and startups taking advantage of this technology to fulfill and streamline a vast list of tasks. And this list will only proceed to grow as the accuracy of the technology continues to improve.