US 12,254,631 B2
Dual-level model for segmentation
Peng Wang, Los Angeles, CA (US); Xueqing Deng, Los Angeles, CA (US); and Xiaochen Lian, Los Angeles, CA (US)
Assigned to Lemon Inc., Grand Cayman (KY)
Filed by Lemon Inc., Grand Cayman (KY)
Filed on Jan. 26, 2022, as Appl. No. 17/585,301.
Prior Publication US 2023/0237662 A1, Jul. 27, 2023
Int. Cl. G06T 7/11 (2017.01); G06T 3/40 (2024.01)
CPC G06T 7/11 (2017.01) [G06T 3/40 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20132 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A method, comprising:
inputting data to a first segmentation network, the input data comprising an image and label information associated with the image, the image captured at nighttime and comprising a plurality of regions;
determining at least one region among the plurality of regions based at least in part on output of the first segmentation network;
cropping the at least one region of the image;
inputting the cropped at least one region to a second segmentation network;
producing a final output based on the output of the first segmentation network and output of the second segmentation network;
the method further comprises:
detecting the at least one region in the image using a region detection network; and
cropping the at least one region based on output of the region detection network, wherein the output of the region detection network comprises at least one bounding box.