Hazen.ai: A KSA Startup Using Deep Learning To Drive Road Safety Hazen.ai is a company that uses computer vision and deep learning to design intelligent traffic analytic software to help reduce road accidents.

Meet Sohaib Khan, distinguished professor turned entrepreneur. Khan is silver-haired, and he has the nurturing smile you might expect from one of your favorite former professors. He brings decades of research experience, first from completing his PhD in the United States, and later from running computer vision research and innovation labs in top universities in both Pakistan and Saudi Arabia. Today, he is the CEO and one of four co-founders of Hazen.ai, a startup building cost-effective artificial intelligence (AI)-driven software solutions that aim to reduce road fatalities. He is part of a new brand of entrepreneurs emerging out of Saudi Arabia, with deep expertise and know-how, and a distinguished career behind him.

He and his team are commercializing computer vision applications. "Commercializing computer vision applications was seemingly implausible just over a decade ago," Khan explains. "It was largely seen as a fringe research science that could not withstand the rigor of practical or commercial applications." What changed in the last decade was several breakthroughs in the research, with seminal research papers published on "unsupervised (machine) learning" and "deep neural networks." At the same time, computing power was increasing exponentially each year, and the hardware had finally reached capabilities that could facilitate machine-learning. Last, but not least, the availability of data -such as a library of images needed to feed computers as a deep learning data set- also played a critical role. Today, we see computer vision applications in autonomous vehicle (AV) startups, or facial recognition software.

So, how did Khan decide to utilize computer vision in road safety applications? "Unfortunately, there are 1.3 million unnecessary road accidents all over the world each year, and a disproportionate number of these deaths are in emerging markets," Khan notes. "We don't call them road accidents at Hazen.ai though. We call them crashes, because they are avoidable. 9.9 times out of ten, it comes down to human error, in the form of bad driving, or texting and driving, or poor traffic design, or faulty roadwork, and even poor signage. All of these elements are addressable if we could just make the observations, and issue immediate recommendations for rectification."

That's where Hazen.ai comes in. Taking his nearly two decades of computer vision research, Sohaib founded Hazen.ai with one of his former Pakistani students and two Saudi co-founders. Hazen.ai has essentially turned a hardware solution, often in the form of speed radars and cameras, with limited capabilities, into a software solution with unlimited potential. Turning hardware into software solutions allows for artificial intelligence to continue leveling up the product performance, adding more complex capabilities overtime, while reducing the cost of upgrades and implementation.

For example, a typical radar system on a four-way intersection in the United States could cost upwards of US$800,000; roughly 25% for the hardware equipment, and 75% for installation, certifications and maintenance. This radar system's only capability today is to ticket drivers for infractions like speeding, or not heeding a red light. In contrast, Hazen.ai's solution seeks to offer an intelligent solution for one-tenth of the cost. "If there are already cameras installed at an intersection, all that is required is to install a small processor next to the existing camera, reducing the cost even further," Khan explains.

Hazen.ai has the ability to monitor and detect different types of dangerous and unsafe behaviors like unsafe lane changes, illegal turns, jumping a red-light or a stop-sign, blocking a pedestrian crossing, not wearing a seat belt, or texting while driving. The software was also recently upgraded to detect safety hazards for pedestrians and cyclists. Despite every country having different traffic rules, the software has been built for easy configuration in just a few clicks. "We've built it so that it's easily scalable across the world with customization capabilities built-in for a smooth onboarding process," he explains.

"We're driven by the million-plus lives we can save each year," Khan says, as he gives the airline industry as an example of a sector that has made air travel statistically safer than driving a car because of the investment, technology, and importance given to safety. The flight crew, traffic control, and aircraft dispatchers all work together to ensure a safe and successful flight. There's a central regulatory authority, hours of training for pilots, and technology deployed regularly in the name of safety. When there is a plane crash, which is incredibly rare, it makes the news. Every plane crash is taken seriously, with diligent investigations. Today, air travel results in 0.07 deaths per one billion miles travelled, compared to 7.28 deaths per one billion miles travelled in a car. That makes air travel more than 100x safer than driving a car. "Or another way to put it: every 24 seconds, someone dies in a car crash, which is equivalent to 15 787 Dreamliners crashing every single day!" Khan says. "This is what motivates us."

Today, Hazen.ai is revenue-generating, and it is currently running pilot programs in several countries including Saudi Arabia, New Zealand, United Kingdom, and United States. The company has also successfully raised pre-seed and seed capital from two angel investors and Wa'ed Ventures. "The team at Wa'ed have been super supportive, and an excellent strategic partner for Hazen.ai," he says. "They understood the importance of this work, and that it requires patient capital." Hazen.ai is now gearing up to raise its Series A soon, and it is keen to meet international and regional strategic investors familiar with deep tech investments.

As road fatalities hit developing countries disproportionately, and with only 0.1% of the world's intersections being tech-enabled to date, Khan and his team believe Hazen.ai is in a unique position to be a key player in the fight for global road safety, and in the fight to save avoidable lives lost. While there are other companies attempting to develop similar solutions in the West, not many seem to have exposure to, or are willing yet, to enter the developing markets that need this life-saving technology the most.

When I ask him if the emergence of autonomous vehicles might make his technology obsolete, he pauses to ponder the question for a moment, and responds. "First of all, a world ruled by autonomous vehicles won't happen overnight, and we need to reduce road fatalities today. But even in a world where AVs were our sole mode of transportation, I don't see Hazen.ai as a plug-the-gap technology, but more of a complimentary technology that will evolve with society and autonomous vehicle technology overtime." Khan also points toward other factors such as roadworks, pedestrian and cyclist behaviors that all need to be taken into account to ensure safe roads in the future, and that would still require a Hazen.ai-type solution. "If anything, I would love to see our technology evolve so that our cameras and software directly speak to AVs on the road to warn them of upcoming dangers in real-time." Now, that would be a safer world indeed, and perhaps a future that is not as far off in the distance as many of us might think.

Learn more about the Saudi Arabia startup ecosystem by checking out the report, The Evolution of Saudi Arabia's Startup Ecosystem 2010-2022.

This article was originally published on Lucidity Insights, a partner of Entrepreneur Middle East in developing special reports on the Middle East and Africa's tech and entrepreneurial ecosystems.