US 12,073,633 B2
Systems and methods for detecting traffic lights of driving lanes using a camera and multiple models
Kun-Hsin Chen, San Francisco, CA (US); Kuan-Hui Lee, San Jose, CA (US); Chao Fang, Sunnyvale, CA (US); and Charles Christopher Ochoa, San Francisco, CA (US)
Assigned to Toyota Research Institute, Inc., Los Altos, CA (US); and Toyota Jidosha Kabushiki Kaisha, Toyota (JP)
Filed by Toyota Research Institute, Inc., Los Altos, CA (US)
Filed on Apr. 22, 2022, as Appl. No. 17/726,939.
Prior Publication US 2023/0343109 A1, Oct. 26, 2023
Int. Cl. G06V 20/58 (2022.01); G06V 10/56 (2022.01); G06V 20/56 (2022.01)
CPC G06V 20/584 (2022.01) [G06V 10/56 (2022.01); G06V 20/588 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A detection system for identifying traffic lights comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the processor to:
estimate, from an image using a first model, depth, orientation information, and location information of the traffic lights relative to a driving lane of a vehicle within a traffic scene;
compute, using a second model, relevancy scores for the traffic lights according to geometric inferences between the depth, the orientation information, and the location information, and the geometric inferences represent the orientation information as an angular relationship between the vehicle within the driving lane and the traffic lights;
assign, using the second model, a primary relevancy score from the relevancy scores for a light of the traffic lights associated with the driving lane according to the depth and the orientation information; and
execute a control task by the vehicle according to the primary relevancy score and a state confidence, computed by the first model, for the light.