首页|Patent Issued for Systems and methods for multi-modal data augmentation for perc eption tasks in autonomous driving (USPTO 12145592)
Patent Issued for Systems and methods for multi-modal data augmentation for perc eption tasks in autonomous driving (USPTO 12145592)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
From the background information supplied by the inventors, news correspondents o btained the followingquote: “In development of data for training machine learni ng models, data collection and labeling is alaborious, costly, and time-consumi ng venture, which may represent a major bottleneck in most currentmachine learn ing pipelines. In particular, recent perception models in self-driving industry have startedto focus on the use of multi-modal data fusion. Typically, the mult i-modal data is associated with oneor more red-green-blue (RGB) cameras and/or one or more light detection and ranging (LiDAR) sensors.This typically requires the annotations of multi-modal data from several sensors to be synchronized, which may be relatively difficult. Furthermore, when collecting real world data fo r annotation, such data istypically limited to real traffic data of a certain t ime period from a few environments (e.g., cities and thelike). Thus, even a wel l-annotated dataset is not able to cover all the scenarios that an autonomouslyoperated vehicle may face.”
NETL
NSTL
