Axon offers technology boost for public safety with in-car Automated License Plate Recognition on Azure

Building a public safety ecosystem

People around the world increasingly turn to technology to play an important role in meeting today’s many pressing challenges, and modern innovations have made previously unavailable, unimagined solutions possible in fields as diverse as climate, medicine, logistics and transportation, entertainment, and finance. At Axon, Director of AI Engineering Kate Puech helps bring solutions to another area that presents opportunities for technological intervention: public safety.

Founded in 1993 and formerly known as TASER International, Axon has a mission-driven focus to protect life, capture truth, and accelerate justice by innovating across technology, training, and software for public safety. The company offers a range of connected products, devices, and services that are designed to make public safety, including law enforcement, safer, more effective, and more efficient for communities and the customers that Axon serves. Kate oversees the team building the infrastructure that the company uses to bring the very latest AI technologies to the mix of products that Axon calls its ecosystem of connected solutions.

One of the latest projects for Kate’s team was to add real-time license plate recognition to the company’s Fleet 3 in-vehicle video solutions. “The system proactively lets you know if it identifies a wanted car and can trigger an alert to take appropriate action,” Kate says. For example, license plates might be associated with an individual of interest, a missing person or AMBER Alert, a suspected stolen vehicle, or a vehicle that was involved in an accident. Axon develops and trains machine learning models, which are automatically downloaded and installed on the in-car devices cameras where the models run.

Meeting challenges using Azure technology

For most AI solutions, providers need the power and resources of the cloud to meet their development, build, and test demands. Axon, too, prefers the efficiency of a cloud-based environment over the substantial and ongoing on-premises capital investment that would otherwise be needed. Kate says, “Microsoft Azure infrastructure, including Azure Machine Learning, helped us develop an in-house automated license plate model that’s deployed across the United States. This isn’t trivial!” Not trivial because real-time analysis of moving license plate images requires prior model training on diverse, highly variable data. “In the United States, each state has different types and styles of plates,” Kate notes. “We have to train the models to deal with a lot of scenarios: night, day, dusk, parking lots, open roads, and extreme angles and speeds.” Axon plans to continue innovating with more solutions in the future. “Automated License Plate Recognition has been very well received by our US customers, and we’re now investing in expansion of the service internationally.”

Kate adds that the Azure platform’s on-demand compute and storage are another key value. “The ability to scale compute resources up and down is critical for innovation speed and cost efficiency because the amount of horsepower we need can vary dramatically depending on the task.” Essential AI features and services in Azure also contribute to an efficient development environment. “We use PyTorch on Azure and Azure compute and storage resources to develop AI responsibly and at scale,” she says. “Thanks to Azure Machine Learning, experiments are reproducible, and the model store and dataset store provide the right primitives to properly version our datasets and models. Azure Machine Learning and its built-in machine learning operations capabilities make agility and cost-efficiency simple.”

Axon’s success is grounded in ethical practices and compliance with mandates wherever its solutions are offered. Azure helps by providing the right primitives, such as Azure Virtual Network and managed identities, to compliantly collect data and train models overseas. Kate appreciates the peace of mind that the company’s chosen cloud platform provides. “We get excellent primitives to develop compliant solutions from Microsoft. That’s a big selling point for us,” she says.

Find out more about Axon on Twitter, Facebook, and LinkedIn.