Data is the “fuel”that drives AI system towards continuous progress, but building high-quality datasets isn’t easy. The part where involve data collecting, cleaning, annotating, and privacy protecting are all challenging. Researchers need to collect targeted data to deal with complex problems faced on different fields to make sure the trained models have robustness and generalization capability. Through using rich datasets, AI system can achieve intelligent decision-making in more complex scenario.

However, since the technology of intelligent driving is not perfect, some functions still have certain risks if self-driving is applied, so people need to keep attention while driving.

Whether the driver is paying attention to driving is a technical problem. Current technologies identify driver’ s attention by tracking their behaviors, such as touching the steering wheel, driver’s eyes, facial expressions, eye blinking speed etc. However, these technologies have some limits. They are not very friendly, for example the driver needs keep hands on the steering wheel. Or they are prone to misjudgment, such as eye blink detection may not recognize whether the driver is looking at the scenery outside the window.

Recently, Google disclosed a patent “VEHICLE OCCUPANT ENGAGEMENT USING THREE-DIMENSIONAL EYE GAZE VECTORS”. According to the description, it can more accurately locate the relative position and specific components in the vehicle that the user is viewing through the 3D gaze vector of the eyes, such as whether the user is looking at the rearview mirror, instrument panel, etc.

It can determine the three-dimensional eye gaze vector based on the occupant’s facial plane. Using the 3D position of the occupant’s head/eyes in the cabin space and the 3D eye gaze vector, it is possible to more accurately determine the position in the 3D cabin space that the user is looking at, such as the rearview mirror, the vehicle’s head unit, the vehicle’s instrument displays, front windshields of vehicles, etc.

This technology can effectively track the driver’s attention and avoid the accidents caused by the driver’s distraction. Nexdata has developed Multi-race — Driver Behavior Collection Data and Driver Behavior Annotation Data to support the large-scale application of driver’s behavior recognition. Nexdata strictly abides by the relevant regulations, and the data is collected with proper data collection authorization agreement and under the ISO 27001 Privacy Information Management System certification and the ISO9001 Quality Management System certification.

Multi-race — Driver Behavior Collection Data

The data includes multiple ages and multiple time periods. The driver behaviors includes Dangerous behavior, fatigue behavior and visual movement behavior. In terms of device, binocular cameras of RGB and infrared channels were applied.

Driver Behavior Annotation Data

The data includes multiple ages, multiple time periods and behaviors (Dangerous behaviors, Fatigue behaviors, Visual movement behaviors). In terms of annotation, 72 facial landmarks (including pupils), face attributes, gesture bounding boxes, seatbelt bounding boxes, pupil landmarks and behavior categories were annotated in the data.

Due to technical limits, current intelligent driving vehicles cannot driver alone without human’s intervention. We believe that the real self-driving vehicles will come in the near future.

End

If you need data services, please feel free to contact us: info@nexdata.ai

While pushing the boundaries of technology, we need to be aware of the potential and importance of data. By streamline the process of datasets collection and annotation, AI technology can better handle various application scenarios. In the future, as datasets are accumulated and optimized, we have reason to believe that AI will bring more innovations in the fields of medication, education and transportation, etc.