US 12,073,632 B2
Structural object detector for hierarchical ontology for traffic light handling
Kun-Hsin Chen, Mountain View, CA (US); Dennis I. Park, Fremont, CA (US); and Jie Li, Los Altos, CA (US)
Assigned to TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed by TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed on May 11, 2020, as Appl. No. 16/872,095.
Prior Publication US 2021/0350152 A1, Nov. 11, 2021
Int. Cl. G06V 20/58 (2022.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 20/00 (2019.01)
CPC G06V 20/584 (2022.01) [G06F 18/2148 (2023.01); G06F 18/217 (2023.01); G06N 20/00 (2019.01)] 17 Claims
OG exemplary drawing
 
1. A method, comprising:
detecting, from a plurality of images, one or more illuminated traffic signals;
labeling each of the plurality of images with one or more labels indicative of one or more characteristics and operating conditions of the one or more illuminated traffic signals in accordance with a hierarchical ontology, wherein the hierarchical ontology comprises a plurality of levels, the plurality of levels comprising:
a first level corresponding to a first bulb state of one or more illuminated traffic signals;
a second level corresponding to a second bulb state of one or more illuminated traffic signals; and
a third level corresponding to a third bulb state of one or more illuminated traffic signals;
validating a label associated with the third level of the hierarchical ontology based on a label associated with one or more of the first level and the second level of the hierarchical ontology; and
training a traffic signal recognition model with the labeled images, wherein the traffic signal recognition model iterates through only those levels of the plurality of levels of the hierarchical ontology, necessary for traffic signal recognition.