Can you believe a security camera can stop crimes before they happen? If you believe, you may wonder how it works. You may image that a camera will send alerts and video footage to a security guard when it captures a robbery. After that, the guard can rush to the robbery spot and call the police to stop the crime. This scenario may sound like a scene in Sci-fi movie, but it does exist in reality now.
Photo courtesy: umbocv June. 3, 2016 Retrieved from https://www.umbocv.com/#camera
The one that build this crime identifying system is not a fortune 500 company but a startup established in 2014, called Umbo CV. Despite being a security startup, they are few companies that have technical teams of computer vision and machine learning. They combine the two technologies to build image recognition system which is the brain of their system.
Photo courtesy: umbocv June. 3, 2016 Retrieved from https://www.umbocv.com/#camera
Up to now, companies that master computer vision and machine learning are leading technology firms like Google, Apple, and Amazon. They mostly apply the technologies in self-driving cars and drones. Umbo CV, on the contrary, aims at security field. With this unique selling point, the startup receives a total of 9.2 million NT dollars investment from venture capital firms such as Wistron and Phison.
In the security industry, most vendors only focus on upgrading the resolution of recording video, i.e. they improve the resolution from 1080p to 4k. For active protection, most systems rely on facial recognized systems to ferret out suspicious people. The other way is dividing the monitoring areas into several blocks to analyze the change and determine if accidents happen. If it happens, the systems will put the symbol of abnormal activity on the screens.
Photo courtesy: cctvcameraworld June. 3, 2016 Retrieved from http://www.cctvcameraworld.com/4k-ip-camera-reviews/
Unlike normal vendors’ systems, Umbo CV’s image recognition system has self-learning and analytical skills. It can detect abnormal activities and learn from them in a short time. If the system detects abnormalities such as invasion, fire, or robbery, the system will instantly notify the responsible units to prevent tragedies. Moreover, the system can recognize more complex things and objects with more data and learning.
Based on the experience of the Co-founder, he knows his system can be the best solution for companies because most companies cannot afford 24/7 security services, especially for countries that have high labor costs. Moreover, security guards rarely stare at monitors because nothing happens for 99% of the time. His system can help companies reduce the maintenance costs and improve the efficiency of security guards.
To further test the market of his products, the Co-founder takes the prototypes to meet clients in the US, Australia, and Dubai. After he receives orders for 40 million NT dollars, he starts to mass-produce his products.
Up to now, the estimated number of surveillance cameras around the world is 3 billion, and only about 1% of them supports AI technology. “Based on my experience, there are over 60 % percent of customers that want to purchase a security system that supports AI.” the co-founder said optimistically.
Reference:
http://www.bnext.com.tw/article/view/id/39062
https://www.umbocv.com/#camera