Human surveillance is not only expensive, but it’s also not always reliable. Especially at night, attention and concentration can wane at times, and even if this does not happen, our perception in the dark is far from flawless. Camera surveillance can provide an answer to this.
Efficiency is key here, because: a system that requires high maintenance or regularly generates false alarms has little added value in a market where margins are under pressure and where an acute shortage of both technical staff and dispatchers to staff the control rooms is.
Video analysis using AI can provide an answer to this. Still, one AI application is not like the other.
What we often see today is that the initial detection is still done with traditional security cameras, and afterwards only a few snapshots are analysed in the cloud by AI. This helps to reduce the number of false alarms, but also implies a large margin of error, and thus the risk of vandalism, burglary, cracking, and theft.Mathias Germeau, operations manager
Watchtower therefore partnered with one of the leading parties in the field of Computer Vision-based AI Video Analytics, which also works for Airbus and the US Army, for example.
"They developed a new and unique AI model for us, which is specially designed for use in temporary security and allows all images to be analysed immediately on location, instead of just a few snapshots. This increases accuracy by up to 98%, making it very difficult to enter somewhere unseen," Germeau continued.
Specifically, the AI solution makes it possible to analyse live video images integrally and in real time, recognising patterns - such as a human silhouette, for example - and anomalies without human intervention.
"By using Deep Learning algorithms trained on synthetic data, we are now able to run, on energy-efficient hardware, a very powerful AI model that can interpret imagery in a way similar to that of the human brain: perceiving events, recognizing patterns, identifying people, distinguishing objects and noting anomalies without human intervention, in a much more sophisticated way than is the case with traditional security cameras," explains Guido Christ, technical manager at Watchtower.
The benefits of that approach are clear:
It is also important to pay attention to the potential risks involved in using this technology. To avoid privacy issues, for instance, it should be strictly ensured that personal data are always collected and stored in accordance with the law.
In addition, with traditional AI algorithms, there is the danger that a lack of varied training data leads to errors and bias. WatchTower therefore works with synthetic training data that does not discriminate based on race, gender, or any other external characteristic, specifically to avoid such biases.
The future of AI video analytics is promising, with further developments in Deep Learning and Computer Vision that will further improve the accuracy and performance of the systems. Moreover, integrations with other technologies, such as the Internet of Things (IoT) and facial recognition, could open new opportunities for security, management, access control, personalised services, and smart city planning. Some examples:
Learn more about Watchtower Security Solutions: