Swedish start-up Devant is using its human-centric synthetic data to help Seeing Machines monitor drivers and vehicle occupants.

Australian firm Seeing Machines is a computer vision technology company that designs AI-powered monitoring systems to improve transport safety. Devant is supporting the training, testing and validation of machine learning to help Seeing Machines improve its driver and occupant monitoring system (DMS and OMS) technology.

Seeing Machines is developing and licensing proprietary eye-tracking technology, underpinned by bespoke AI that monitors driver attentiveness, for automotive manufacturers. Its technology is powered by machine-learning networks that need to be trained and validated using large sets of high-quality data.

This collaboration brings together Seeing Machines’ access to human behavioural insights with Devant’s rendering capabilities to accelerate the development and training of its machine-learning networks that underpin its DMS and OMS.

This is part of a larger digital engineering initiative to improve development efficiency to meet the growing demand for more in-cabin, driver and occupant monitoring capabilities as more transport safety regulation deadlines approach. The incorporation of computer-generated synthetic data follows Seeing Machines’ research and development to consider all attribute and noise factors such as human appearance and behavioural variations, looking at the optical, electronic and vehicle cabin environment.

The goal is to improve the real-world fidelity of data used to augment the development, training and validation of its in-cabin AI vision software.

Seeing Machines’ scientific approach to validate what makes a computer-generated human real enough in practice, prosecuted on a feature-by-feature basis for DMS and OMS, combined with Devant’s capability and focus to solving the critical in-cabin human synthesis problems, is unlocking a source of synthetic data for accelerated development of in-cabin sensing technology. It is also satisfying Seeing Machines’ requirements to achieve better safety outcomes.

“With a wave of new transport safety regulations coming into force around the world, it is increasingly vital for the advanced machine-learning networks that underpin our AI-driven technology to access reliable data of the highest quality, accuracy and realism,” said Paul McGlone, CEO of Seeing Machines. “Our mission to get everyone home safely has never been more relevant. That is why we have partnered with Devant, a leader in its field, to help deliver on our promise of providing our customers with bespoke, mission-critical options that enhance driver safety.”

Devant can capture and translate the subtlest movements and human behaviour, producing a granularity of detail that has, until now, been impossible to simulate. This approach to data creation should allow machine learning to cover a broader range of human activity.

“We pride ourselves on delivering best in class synthetic data, combining unparalleled 3D human animation, computer graphics and engineering to deliver millions of images and animations,” said Richard Bremer, CEO of Devant. “We are pleased to be partnering with an innovative market leader like Seeing Machines, which is delivering a step change in transport safety by harnessing its class leading DMS technology.”

Seeing Machines uses machine-vision technology to monitor driver behaviour under a full spectrum of lighting conditions, including through sunglasses. These data are processed to interpret driver attention state, focus, drowsiness and impairment levels to provide critical inputs in real time to advanced driver assistance systems (ADAS), as well as to vehicle cockpit, comfort and convenience systems.

With DMS set to become mandatory around the world this decade, Seeing Machines hopes to increase the close to 900,000 cars on the road today using its safety technology, underpinned by 15 expanding automotive programmes for ten car makers.

Seeing Machines was founded in 2000 and is headquartered in Australia.