With criminal activities on the rise and terrorist threats more imminent than ever, city and law enforcement officials rely on video surveillance to crack down on crime. But sifting through piles upon piles of video data manually can be a daunting task. This is where smart search analytics tools can come in handy.
With criminal activities on the rise and terrorist threats more imminent than ever, city and law enforcement officials rely on video surveillance to crack down on crime. But sifting through piles upon piles of video data manually can be a daunting task. This is where smart search analytics tools can come in handy.
“In this day and age, more cameras mean more data, but not necessarily better information. Search uses a wealth of data to find the information you need,” said Mahesh Saptharishi, CTO of Avigilon. “Advanced search functions are increasingly important because they help detect people and objects of interest quickly, allowing operators to intelligently comb through video data to produce real world insights.”
“For a major incident it can take many hundreds of man hours manually reviewing footage. In this instance, where time is critical to the success of an investigation, the ability to automate much of the process saves huge amounts of time and resource, improving the likelihood of a successful resolution,” said Erez Goldstein, Senior Product Marketing Manager at Qognify.
In fact, smart search isn’t a mere post-event investigation tool. More and more, it can play a critical role in event management as well. “Say you have an ongoing situation happening right now, and you want to evacuate the building and make sure it is empty. You can use our smart search capability to check if the building is empty and also try to locate the suspect to see if he is in the building right now,” said Zvika Ashani, CTO of Agent Video Intelligence (Agent Vi).
How it's done
Most smart search analytics in the market work in the same way: a search is performed on the video archives, and all relevant data based on certain criteria are retrieved. “Search by keyword or similar object is important, for example clicking to highlight an object and saying 'find me all instances of video that contains something like this,’” said Ben White, CTO of VCA Technology.
Smart search can be used to find not only people of interest but also vehicles or other objects. “Our latest update is that we have added face and license plate search to MomentQuest. The license plate search works quite similarly: it detects a license plate in the field of view or in video footage from one or more cameras. The algorithm involves advanced heuristic methods (such as substitution of similar-looking letters or numbers) to identify as many potential matches as possible,” said Alan Ataev, Global Sales Director at Axxonsoft.
Differentiators
Since smart search software has become more common, vendors and developers try to make theirs unique and stand out. The following are some unique selling points that the vendors claim to set their products apart.
Qognify: One unique feature of Qognify’s Suspect Search is the use of a human “avatar” – in addition to photos or video footage – as a reference for the search. “It is a very common use case that a suspect needs to be located based on a phone call description, so an additional unique and useful capability (of our product) is to be able to trigger a search based on an image taken from outside the system, the most common scenario being an image sent from a mobile phone camera,” Goldstein said.
Avigilon: One of the key differentiators of Avigilon’s Appearance Search is that video analytics is distributed across the solution. “This enables us to scale dramatically across many cameras without the need for additional resources, effectively reducing the total cost of ownership,” Saptharishi said. “Our technology works with our self-learning video analytics and our 1 to 16 megapixel cameras. Using our technology, security personnel can answer ‘who,’ ‘what’ and ‘when’ across an entire site – not just a single camera – to determine how to act.
BriefCam: BriefCam's Video Synopsis technology comes with a special feature whereby multiple objects, events and activities that have occurred at different times during the day are presented simultaneously on the screen, essentially summarizing hours of video recordings into a short “brief” that takes only minutes to review.
AgentVi: Its technology comes with a unique feature whereby the processing is shared between the camera and the server. “In our case, we have a unique architecture where some of our code is embedded in the camera, but we also have a server,” Ashani said. “At the end of the day, for us it's the server that's providing the result, but there is also some processing in the edge, and this enables us to have much higher accuracy and a much richer feature set than would be possible if everything was embedded in the camera.”