Detecting and Preventing Crimes Through AI and Machine Learning
Sun, April 18, 2021

Detecting and Preventing Crimes Through AI and Machine Learning

Every day, we see news of robbery, murder, and other criminal acts / Photo by: TheaDesign via Shutterstock


Every day, we see news of robbery, murder, and other criminal acts. The World Population Review, an independent organization that aims to make data more accessible through graphs, charts, analysis, and visualizations, reported that the country with the highest crime rate in the world is Venezuela with 84.86. It is attributed to the country's poor political and economic environment. 

The country currently has the second-highest murder rate in the world, with violent crimes like murder and kidnapping increasing every year. Some of the countries with the highest crime rates across the world include Papua New Guinea (80.26), South Africa (77.02), Honduras (75.84), and Afghanistan (73.26). Despite years of effort to prevent crimes from happening, they remain part of our society.

Detecting specific patterns of crime and criminal behavior is extremely challenging. According to Wired, a monthly American magazine that focuses on how emerging technologies affect culture, the economy, and politics, countless hours are usually required to sift through data to discover new patterns or to determine whether a crime fits into a known pattern. The information would only be useful once a pattern is detected. It can be used to predict, anticipate, and prevent crime.

Unfortunately, detecting specific patterns of crime largely remains a manual task. As of now, the tools that police are using have no capability to fund specific patterns of crime. What the authorities can do is estimate crime levels. Aside from that, analysts usually need to review crime reports and compare them to past crimes, which usually takes forever. Fortunately, artificial intelligence and machine learning can help.

AI and ML for Crime Detection

Recently, the Kolkata Police in India announced that they would be using more high-tech cameras with AI and face recognition facility in their city to detect crimes. This is after an AI-powered CCTV camera had helped the police identify and punish a citizen for spitting and defacing public property in Gujarat. According to Analytics India Magazine, which covers technological progress in the space of analytics, artificial intelligence, data science & big data in India, the cameras can identify a citizen’s body movements, interpret them, and understand that the person was breaking the law.

Aside from that, the country’s Special Investigation Team used AI-based algorithms last year to probe the murder of journalist-activist Gauri Lankesh. The algorithms were able to gather CCTV footage and identify the people who matched the physical description given by eyewitnesses to the gruesome murder. Thus, it’s not surprising that the Indian Police has started becoming more interested in crime analytics using big data. This involves storing and analyzing huge volumes of various types of data in real-time.

Recently, the Kolkata Police in India announced that they would be using more high-tech cameras with AI and face recognition facility in their city to detect crimes / Photo by: Saikat Paul via Shutterstock


Many companies are also developing tools to detect crime using AI and machine learning. For instance, software company Predpol used these technologies to predict when and where new crimes are most likely to occur. The algorithm they are using can predict where future crimes will happen with the help of the historical data and observing where recent crimes took place. Their system also uses a technique called real-time epidemic-type aftershock sequence crime forecasting where it can highlight possible hotspots the police should consider patrolling more heavily. As of now, the system is being used in several American cities such as Los Angeles. 

Cloud Walk Technology, a Chinese facial technology, is also planning to use facial recognition and gait analysis technology to predict who will commit a crime. This advanced AI can detect if there are any suspicious changes in their behavior or unusual movements. For instance, if an individual seems to be walking back and forth in a certain area over and over, this might indicate that they might end up pickpocketing or partaking in another crime.

Predictive Policing

Predictive policing refers to the usage of predictive and mathematical analytics to identify likely targets for police intervention and prevent crime. Also, it can solve past crimes by making statistical predictions. According to CIO, an online site that delivers the latest tech news, analysis, how-to, blogs, and videos for IT professionals, it revolves around the three most important data points: the location of the crime, date and time of the crime, and type of crime. 

Machine learning can greatly help in predictive policing because crime patterns can be automatically identified. The police could then stop the crime from happening. Without such tools, it would take weeks or years of sifting through a database to discover a pattern. Reports show that predictive policing has been succeeding for the past decade since major crime trends have been spiraling downward. 

However, not everyone agrees with this. Andrew Ferguson, a professor of law at the University of the District of Columbia and author of the book titled “The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement,” stated that there’s a real danger with any kind of data-driven policing because this tends to forget that there are human beings on both sides of the equation. 

“Officers need to be able to translate these ideas that suggest different neighborhoods have different threat scores. And, focusing on the numbers instead of the human being in front of you changes your relationship to them,” Ferguson said.

Nonetheless, AI and machine learning continue to be an enormous success in crime detection and prevention. These technologies make authorities'  jobs easier and their actions more effective.