专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明公开

EP3933439A1 LOCALIZATION METHOD AND LOCALIZATION DEVICE 审中-实审
公开(公告)号：EP3933439A1
公开(公告)日：2022-01-05
申请号：EP19923505.2
申请日：2019-04-03
申请人： Huawei Technologies Co., Ltd.
发明人： ZHANG, Sunpei , FENG, Yuan , LI, Wei , ZHOU, Peng
IPC分类号： G01S17/06 , G01S17/93
摘要： Embodiments of this application disclose a positioning method and apparatus. The positioning method and apparatus relate to the artificial intelligence field, and specifically to the self-driving field. The positioning method includes: clustering points in a first point cloud through multi-clustering to obtain a target point cloud, where the target point cloud is a point cloud that represents a feature of a target object, and the first point cloud includes the target point cloud and a point cloud that represents a feature of an interfering object; and determining a position of the target object based on the target point cloud. In the embodiments of this application, the point cloud that represents the feature of the target object is selected through multi-clustering, to achieve accurate and highly real-time positioning of an obstacle with a relatively small quantity of reflection points.

2. 发明公开

EP4075343A1 DEVICE AND METHOD FOR REALIZING DATA SYNCHRONIZATION IN NEURAL NETWORK INFERENCE 审中-公开
公开(公告)号：EP4075343A1
公开(公告)日：2022-10-19
申请号：EP19958452.5
申请日：2019-12-31
申请人： Huawei Technologies Co., Ltd.
发明人： WANG, Yanyan , FENG, Yuan , WU, Zuguang , ZHOU, Peng
IPC分类号： G06N3/08
摘要： This application provides an apparatus and a method for implementing data synchronization during neural network inference, relates to an artificial intelligence (Artificial Intelligence, AI) field, and specifically relates to neural network inference technologies. The apparatus includes: a memory, configured to store a first feature map; and a neural-network processing unit NPU, configured to: obtain the first feature map from the memory, where the first feature map includes M blocks, and M is a positive integer; separately perform, in an asynchronous manner, inference computation at at least two layers of a neural network model on the M blocks to obtain M inference results, where the asynchronous manner means that data synchronization is not performed on an intermediate result obtained after inference computation at one layer of the neural network model is performed on each block, and inference computation at a next layer continues to be performed on the intermediate result; and perform data synchronization on the M inference results to obtain synchronized data. Because the NPU performs data synchronization only after completing inference computation at the at least two layers of the neural network model, a quantity of data synchronization times in a neural network inference process is relatively small, and less data migration overheads are generated.

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式