Taming traffic woes

Jaipur, the capital of the northern Indian state of Rajasthan, is known for its vibrant culture and stunning architecture. Surrounded by a great wall, with hills to the north, the city and its sites — such as the Amber Fort and the pink sandstone City Palace — draw visitors from around the world.

Yet despite its nearly 300-year history, Jaipur has been something of a modern marvel from the start. The city’s design was intentional; nine distinct sections mirror the Hindu belief in nine sections of the universe. Streets follow a distinct grid pattern, with the walled city’s road network leading to well-defined public areas.

In keeping with its heritage, Jaipur was also one of a handful of municipalities to be designated as a smart city within Rajasthan, with a key initiative focusing on the development of a “smart road” equipped with CCTV (closed-circuit TV) cameras, Wi-Fi, and sensors to ensure pedestrian safety.

Just last month, Jaipur city officials pushed forward with plans to upgrade the city’s CCTV surveillance cameras to a higher-range system powered by artificial intelligence (AI). Once implemented, the system will be capable of identifying 11 traffic violations, from high-beam infractions to wrong-way driving, and communicating that information to a central command center.

The situation in Jaipur may be a glimpse into what’s to come in municipalities around the world.

As we learn in this edition’s cover story, “Vision Systems Regulate Traffic, Improve Safety,” transportation agencies within the U.S., for instance, are looking to the Southwest Research Institute to use deep learning to identify changes in visibility related to inclement weather.

Today, this information is relayed to someone who verifies it and then may activate electronic signage to warn drivers or take other steps. But increasingly, the role of humans in this process is likely to diminish. In the not-too-distant future, deep learning will be used to assess the number of standing vehicles and then trigger further action or, as in the case of Jaipur, spot wrong-way drivers. Click herefor the full story.

Also on the topic of deep learning, Artemis Vision’s Tom Brennan offers practical advice for those investing in software packages. Users are buying a very powerful “applied statistics engine”; therefore, to get the most out of any product, he says, prospective buyers should conduct thorough tests ahead of time using real-world examples. Brennan reveals how in “Putting Deep Learning to the Test”.

Finally, if you’re curious about where the global machine vision market is headed in 2020, don’t miss this edition’s “Field of View” by Ronald Mueller, which takes a closer look at the U.S.-China trade war and examines the impact of a stagnating semiconductor and automotive market.

Michael Wheeler, Editor-in-Chief