US 11,869,247 B2
Perception data detection method and apparatus
Dan Li, Beijing (CN); Jianping Li, Beijing (CN); and Qian Chen, Beijing (CN)
Assigned to APOLLO INTELLIGENT CONNECTIVITY (BEIJING) TECHNOLOGY CO., LTD., Beijing (CN)
Filed by Apollo Intelligent Connectivity (Beijing) Technology Co., Ltd., Beijing (CN)
Filed on Sep. 9, 2021, as Appl. No. 17/471,022.
Claims priority of application No. 202011552344.5 (CN), filed on Dec. 24, 2020.
Prior Publication US 2021/0406564 A1, Dec. 30, 2021
Int. Cl. G06V 20/58 (2022.01); G06F 18/22 (2023.01); G06F 18/2113 (2023.01)
CPC G06V 20/58 (2022.01) [G06F 18/2113 (2023.01); G06F 18/22 (2023.01); G06V 2201/07 (2022.01)] 13 Claims
OG exemplary drawing
 
1. A perception data detection method, comprising:
acquiring labeled data and perception data, wherein the labeled data comprises a labeled position and a labeled type of at least one first obstacle, and the perception data comprises a perception position and a perception type of at least one second obstacle;
performing, according to a negative sample in the labeled data, rough filtering on the second obstacle in the perception data to obtain a third obstacle remaining after the rough filtering, wherein the negative sample is a non-physical obstacle;
determining a matching relationship between the third obstacle and the negative sample, and determining the negative sample having the matching relationship with the third obstacle as a target obstacle; and
determining a detection result of the perception data, according to a number of the target obstacle and a number of the first obstacle;
wherein performing, according to the negative sample in the labeled data, the rough filtering on the second obstacle in the perception data to obtain the third obstacle remaining after the rough filtering comprises:
calculating, according to the labeled position of each negative sample and the perception position of each second obstacle, a distance between each negative sample and each second obstacle; and
retaining the second obstacle whose distance from the negative sample is less than or equal to a preset distance to obtain the third obstacle remaining after the rough filtering.