Machine learning models archive excellent results when detecting objects in images. LiDAR sensors provide the automotive industry with the possibility to localize objects in 3-D, especially for autonomous driving or advanced driver assistance systems (ADAS). The task is to provide a 3-D tagging solution for driving situations that were recorded with LiDAR sensors and front facing cameras. This allows validation of obstacle detection sensors and provides statistics about encountered driving situations.
We implemented a solution that uses deep neural networks to fuse sensor input in bird’s eye view and camera input via convolutional neural networks. Bird’s eye view and camera object proposals are merged and provide a 3-D bounding box detection of street scene objects like cars, pedestrians, and cyclists.
Results and Impact
Object boundaries are detected within an accuracy of 20cm and thus provide reliable tags to automatically gather statistics about the environment. A labor intensive manual tagging process could be avoided.
We provide our customers with innovative solutions build on cutting-edge technologies from artificial intelligence and machine learning.
13355 Berlin, Germany
+49 151 19569392
Since 1996 the P3 group represents innovative Engineering services in the areas automotive, telecommunications, aviation, energy and public sector. The foundation are 3,900 highly qualified and motivated employees at over 40 locations in 17 countries.