Could the prospects for video surveillance
Posted: Wed Feb 12, 2025 10:10 am
M.G.: No, everything is fine with the algorithms and processor power. But, I repeat, scenes that are really complex in terms of lighting cannot be processed today by any AI algorithms: there is simply not enough data for analytics in these cases. Imagine what will happen if you trust unreliable analytics in a system with several thousand video cameras (and in practice it is unreliable). The operator will simply get bogged down in processing false positives!
PC Week : in behavioral analysis and object recognition be brighter?
M.G.: Today, behavioral analysis and object nepal whatsapp data are only a dream. It is difficult to distinguish one object from another using AI. For example, distinguishing a suitcase at the airport that was left for a while or lost by the owner from a terrorist’s stash. And if the suitcase is loaded onto a trolley, it is even more difficult.
In addition to what has been said, behavioral analytics is complicated by changes in the behavior of the observed objects: yesterday, a line at a store window was the norm, today it is a reason for the administration to think about the efficiency of the sales assistants; yesterday, a visitor who held a package of goods in his hands could be considered a potential thief, today it may be a health-conscious buyer who carefully studies the composition of the product...
Only a human analyst observing what is happening in dynamics is able to make the right decision. Unfortunately, reality today does not allow us to rely on AI, machine learning, etc. in video surveillance.
PC Week : How do you rate the security of video surveillance systems against cyber attacks today? Will IP video cameras still remain "active" members of cybercriminal botnets?
PC Week : in behavioral analysis and object recognition be brighter?
M.G.: Today, behavioral analysis and object nepal whatsapp data are only a dream. It is difficult to distinguish one object from another using AI. For example, distinguishing a suitcase at the airport that was left for a while or lost by the owner from a terrorist’s stash. And if the suitcase is loaded onto a trolley, it is even more difficult.
In addition to what has been said, behavioral analytics is complicated by changes in the behavior of the observed objects: yesterday, a line at a store window was the norm, today it is a reason for the administration to think about the efficiency of the sales assistants; yesterday, a visitor who held a package of goods in his hands could be considered a potential thief, today it may be a health-conscious buyer who carefully studies the composition of the product...
Only a human analyst observing what is happening in dynamics is able to make the right decision. Unfortunately, reality today does not allow us to rely on AI, machine learning, etc. in video surveillance.
PC Week : How do you rate the security of video surveillance systems against cyber attacks today? Will IP video cameras still remain "active" members of cybercriminal botnets?