One of the most discussed technologies of this time, artificial intelligence (AI) has entered the public safety sphere and is only predicted to grow its role.

“[AI] gives us superpowers and allows us to do things that will never be feasible for humans to do on their own,” said Noah Spitzer-Williams, senior principal product manager at Axon. “It allows us to process and chew through mountains of evidence and opens the door for some incredible ideas we have to keep officers safe.”¹

A powerful tool that reduces an officer’s time spent searching through large amounts of data and records, AI has infiltrated nearly every aspect of policing—from traffic safety to investigations to personnel and evidence management.

Facial Recognition

Biometrics has played a key role in modern policing since the 1900s with advances in ID methods such as fingerprints and DNA. Now, AI research and development is taking law enforcement biometrics to the next step.

For over 20 years, a Vancouver-based company has been proving that there are countless possibilities for AI in facial recognition software. Each of Face Forensics’ product updates have been driven by the needs of the consumer. One of the latest versions was developed for the International Red Cross in Paris who was working to identify drowned North African migrants. This extension of the f2 facial recognition software had been optimized to match scars, distinguishing marks, and tattoos in images that are not high quality, opening the door for partial recognition.

“With face recognition, it’s not like [fingerprint matching] because it’s a noncontact biometric,” explained Iain Drummond, chief executive officer of Face Forensics. “It’s very dependent on things like expressions, lighting, resolution, distance… and that complicates that whole issue.”² Face Forensics’ solution? AI.

With the goal of trying to make the recognition software as generally applicable as possible when it comes to physical appearance variations without generating false positives, the team continued to expand its capabilities. Often utilized for human trafficking cases, the partial recognition software can connect to existing databases in order to match against an agency’s collected data. Key features such as the shape of the subject’s eyes or cheekbones are encoded to create a unique digital signature. This process is completed in mere seconds, and any matches the system finds will be shown as a side-by-side comparison. The investigators will then be able to determine whether a match has been made.

Vehicle Identification

The sheer volume of traffic in many urban areas can make traffic enforcement, vehicle identification, and driver identification extremely challenging. However, AI can help in this area, as well.

Acknowledging the complexities of equipping vehicles with automated license plate readers, plus speed and surveillance cameras, Ekin sought to include them all in one solution: the Ekin Patrol G2. “To us, the solution was clear since all law enforcement vehicles have one thing in common,” said the president of Ekin America, Jerry Diaz.³ The team recognized that every police vehicle had a lightbar, so they decided to replace the traditional light model with an AI-powered “smart” one.

The plug-and-play solution is centrally managed by Ekin Red Eagle OS, streamlining the communication between all deployed systems. Because it is equipped with Ekin Face, the lightbar captures images from a 360-degree patrol view and compares them with an agency’s database, enabling officers to identify suspects more accurately and efficiently.⁴

It also features an automatic license plate recognition function and autonomously performs speed detection and parking management for up to seven lanes of traffic. “[With the lightbar,] police officers can focus on tasks that require human interaction while relying on AI technology to collect data, detect infractions, and alert them of stolen or wanted vehicles,” said Diaz.

Speech Recognition

AI has also made its way into speech recognition interfaces.

Based out of the Czech Republic, Phonexia has developed a voice biometric speaker recognition technology that enables officers to detect a person of interest’s voice, regardless of the words, language, or dialect spoken. “Law enforcement operators often had to manually listen to audio recordings to discover and confirm actionable intel,” said Pavel Jiřík, content strategist at Phonexia. “However, analysts were able to listen to only a fraction of potential recordings of interest.”⁵

The Phonexia Orbis Investigator is a response to the increasing need in law enforcement for intuitive software that can analyze large quantities of audio instantly. The AI-powered analysis can be completed with only a referential voice of a suspect or through a speaker clustering functionality that groups identical voices in audio recording automatically. An insights customization function also allows an analyst to define key words that are most important in the case, which are identified on the intuitive platform when spoken. If audio files contain metadata, such as phone numbers, the Orbis Investigator generates a link analysis to visualize the interactions between the detected voices. This could lead to the discovery of new persons of interest or detect suspicious activity of known suspects.

Ramping up their investment in AI, Axon is creating a new area of research and development called Axon AI. With the intention to have officers spending more time in the field rather than processing data and writing reports, Axon introduced Auto-Transcribe. “The Auto-Transcribe toolset is an instrumental piece of technology in the Axon Ecosystem that allows officers, supervisors, detectives, internal affairs, and attorneys to get right to the key parts of the evidence without wasting time or missing key evidence as they scroll through hours of footage,” said Spitzer-Williams.⁶

As part of the Axon Ecosystem, Auto-Transcribe quickly and accurately generates transcripts of video and audio evidence. Its Fast Evidence Review Assistant function allows officers to search for keywords in a time-stamped transcript, which will direct them to the point in the video in which that topic is discussed. The Transcription Assistant editing tool even helps users create court transcripts within the platform.

Law enforcement is a human-facing job. However, there are many aspects that have taken officers’ focus from those interactions. “In the long term, AI will drive greater efficiency to all aspects of public safety software, reducing paperwork and enabling officers to focus on what matters,” said Spitzer-Williams. AI may be the solution in having officers performing their daily routine in the public and engaging with community members without worrying about time-consuming tasks that take them off the streets.

Notes: 

¹Noah Spitzer-Williams (senior principal product manager, Axon), email interview, September 29, 2022. 

²Iain Drummond (chief executive officer, Face Forensics), phone interview, September 26, 2022. 

³Jerry Diaz (president, Ekin America), email interview, September 28, 2022. 

⁴Ekin, “Ekin Patrol G2.”  

⁵Pavel Jiřík (content strategist, Phonexia) email interview, September 27, 2022. 

⁶Noah Spitzer-Williams, email interview. 

Please cite as:

“Enhancing Recognition with AI,” Product Feature, Police Chief (November 2022): 86–88.