The long-term goal of autonomous driving will require a detailed understanding of complex traffic scenes, in particular the state and possibly intentions of other traffic participants, the most prominent one being non-ego vehicles. An intermediate step and cornerstone to this understanding is a precise estimate of a vehicle's pose and category as well as potential other cues that facilitate predicting its future behavior. The output representations of current state-of-the-art computer vision algorithms, e.g. detection or semantic segmentation, hold little pose information and generally do not straightforwardly allow for any vehicle state analysis. In this work we focus on new vehicle representations that can be learned by semantic segmentation algorithms. We present three vehicle fragmentation levels that divide a vehicle into part areas based on a mixture of material and function, jointly aiming at vehicle state analysis and embedding pose information. To avoid expensive manual labeling we annotate the representations on CAD models to feed a semi-automated labeling pipeline, requiring 3D bounding box labels and a calibrated camera. We evaluate the fine-grained segmentations on two publicly available datasets demonstrating that current state-of-the-art convolutional networks are capable of learning such complex representations and can, hence, exceed straightforward appearance learning.
Fine-Grained Vehicle Representations for Autonomous Driving
01.11.2018
1254681 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
AUTONOMOUS DRIVING VEHICLE, AND PROGRAM FOR AUTONOMOUS DRIVING VEHICLE
Europäisches Patentamt | 2021
|AUTONOMOUS DRIVING VEHICLE AND AUTONOMOUS DRIVING METHOD
Europäisches Patentamt | 2022
|Autonomous driving system and autonomous driving vehicle
Europäisches Patentamt | 2020
|AUTONOMOUS DRIVING SYSTEM AND AUTONOMOUS DRIVING VEHICLE
Europäisches Patentamt | 2019