Computer vision to achieve object detection, tracking and recognition will become hugely important in future Advanced Driving Assistance Systems (ADAS). A variety of smart sensors has become available, including radar, lidar, ultra sound, and video cameras that monitor the environment around the vehicle. Camera-based systems have immensely progressed in recent years and will be an indispensable part of ADAS. Cameras are now cheaper, smaller, and of higher quality than ever before. Concurrently, computing power has dramatically increased. For that reason, NXP started an investigation into the principles, capabilities and future requirements of smart camera systems, being able to detect and track objects with a large degree of accuracy and confidence. In this project, insights about the feature extraction methods for object detection and tracking in automotive domain are gained. Various methods are compared with respect to their suitability for the major use cases, namely, vehicles, pedestrians, traffic signs, and lanes.
|Award date||1 Oct 2014|
|Place of Publication||Eindhoven|
|Publication status||Published - 1 Oct 2014|