SeeQuestor - Video Processing and Analysis Software

SeeQuestor is a fully integrated system designed to radically improve the speed and capacity of handling video footage by law enforcement agencies.

The system can deliver results up to 100 times quicker than current technology and methods by combining the huge processing power of a supercomputer with intelligent software and cutting-edge graphics processors to aid investigations.

SeeQuestor delivers a toolkit of functionality and efficiency improvements that have been designed to enhance the skills of human analysts and investigators in order to enable them to focus their time on other areas of the investigation, reducing precious man hours. The platform is already being used in real cases by law enforcement agencies on four continents.

Henry Hyde-Thomson, CEO of SeeQuestor, said: "For the first time, compute power and graphics capabilities have developed to the extent that we can completely change how we can look at video. Instead of playing video sequentially to look for something interesting, we can pre-index videos with all of the movements, faces and people present; and then find the person or incident of interest many times quicker than ever before."

SeeQuestor is the result of decadesof research & development with a highly skilled software development team and some of the world's leading experts in computer vision and deep learning. It was designed with input from the Metropolitan Police and British Transport Police and has been tested on real cases with police forces from around the world.

"SeeQuestor was created to meet a desperate need in law enforcement: the need to process and analyse thousands of hours of video footage, often of poor quality or unreadable format, and then analyse that footage to fast-track the identification of persons of interest," continued Hyde-Thomson. "It has taken years to get to today where, for the first time, compute power and graphics processing have been sophisticated enough to deliver the required performance to bring SeeQuestor to life."

Created in response to law enforcement challenges

SeeQuestor replaces outdated tools and technology that are recognised as unfit for purpose, at a time when the security environment is becoming ever more challenging.

"The current technology is quite simply not up to scratch," said Detective Chief Inspector Mick Neville, Head of Central Forensic Image Team, New Scotland Yard. He explained: "When the police see enormous amounts of video, that has to be viewed manually and normally it takes one person one hour to review - it's almost an impossibility to do that much. A system that will automate this and quickly scan the footage, it may give us the links and major leads that we can focus on.

"Anything that makes man and machine work together in perfect harmony to get us to identify more criminals and find more victims would be a wonderful tool for the police."

A College of Policing report* in January 2015 noted: "As resource levels fall, there is a risk, as indicated by the HMIC, that the remaining resource time is taken up dealing with reactive demand and less resource remains for preventive work and discretionary activity...including preventative patrolling, community meetings and intelligence gathering. This type of activity which helps the police to understand what is causing high volume offending or problems in hotspots, and come up with specific solutions - often in partnership with others - allows the police to drive down crime."

CCTV is captured in thousands of different formats on a multitude of platforms and systems and after an incident or event, investigators collect hundreds or thousands of hours of footage (100,000 hours for the London riots in August 2011), much of which they cannot access or view easily.

Without standardised formatting, vital evidence may not be accessible to those who need to view it. It is estimated that CCTV footage was recovered for only 16 per cent of crimes.**

Hyde-Thomson explained: "Investigators are often searching for a needle in a haystack. By filtering out unnecessary footage that does not contain a person of interest or activity in a defined location, investigators can quickly identify sections of footage which do contain key evidence."

Cutting-edge functionalities

SeeQuestor has designed a number of core functionalities that significantly reduce the time that an investigator spends reviewing footage. These include: person recognition, face detection and attribute search; geo location information on a dynamic map; and a motion detection filter.

Each functionality will have a significant saving on analysts' time: 24 hours of footage could be indexed in as little as 15 minutes. As an integrated platform, SeeQuestor delivers results that are "unthinkable" with current conventional methods.

Detective Chad Brink of the City of Phoenix Police Department explained how the results achieved with SeeQuestor had delivered a result that was "an unthinkable achievement with technology currently used by law enforcement."

Brink explained: "A case which had gathered footage from more than 30 cameras running 24/7, over five months, would have taken me 15 years to review on my own. Even by just selecting seven core key cameras and a more restricted time range to focus on, a small team would have taken around two years to do an initial viewing of all the footage.

"By utilising SeeQuestor, I was able to review the footage in just four days. The software and workflow management platform also meant that I could view the same scene from more than one angle simultaneously, enabling me to gather vital evidence."

Supercomputer specifics

The SeeQuestor Model 10 and Model 20 are affordable supercomputers built for law enforcement agencies that need to process 50,000 or 100,000 hours of video footage every year. Because all videos are pre-indexed, video forensic teams can process up to ten times more footage than they do at present.

The SeeQuestor Model 20 has a compute power of 60 TeraFLOPS (almost as much as the compute power of the IBM Blue Gene, the world's most powerful supercomputer in 2004).

The processing power of SeeQuestor enables thousands of hours of video to be imported, converted to MPEG4 H.264 format, and then analysed to find all the people. As a result, an investigation team can process thousands of hours of video in a few days of analyst time to find people of interest or the evidence required to progress a criminal investigation.

Human rights and data privacy

The consideration of civil liberties and human rights has been an integral part of the design of SeeQuestor***. Professor Tom Sorell, SeeQuestor's Chief Ethics Officer, stated that the company would not sell to organisations or countries that did not act in accordance with international standards of human rights.

"We look at Amnesty and Freedom House reports for the jurisdictions where the technology might be used," he explained. "If a country to which the technology could be sold registers as problematic in those rating systems, we would look very specifically at the institutions that would be using the equipment, and whether it posed a serious risk of being used against people exercising their human rights."