US 11,868,889 B2
Object detection in images
Zhe Lin, Fremont, CA (US); Xiaohui Shen, San Jose, CA (US); Mingyang Ling, San Jose, CA (US); Jianming Zhang, Campbell, CA (US); and Jason Wen Yong Kuen, San Jose, CA (US)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Jan. 31, 2022, as Appl. No. 17/588,516.
Application 17/588,516 is a continuation of application No. 16/874,114, filed on May 14, 2020, granted, now 11,256,918.
Application 16/874,114 is a continuation of application No. 16/189,805, filed on Nov. 13, 2018, granted, now 10,755,099, issued on Aug. 25, 2020.
Prior Publication US 2022/0157054 A1, May 19, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/08 (2023.01); G06N 3/04 (2023.01); G06V 20/20 (2022.01); G06V 20/64 (2022.01); G06V 10/82 (2022.01); G06V 20/10 (2022.01); G06F 18/214 (2023.01); G06V 10/764 (2022.01); G06V 10/44 (2022.01)
CPC G06N 3/08 (2013.01) [G06F 18/214 (2023.01); G06N 3/04 (2013.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 20/10 (2022.01); G06V 20/20 (2022.01); G06V 20/64 (2022.01)] 20 Claims
OG exemplary drawing
 
1. In a digital medium environment for detecting objects in digital images, a method implemented by a computing device, the method comprising:
obtaining an input digital image that depicts a scene with image objects;
detecting a particular image object depicted in the input digital image with an object detection network that is untrained to detect the particular image object, the object detection network trained by concept conditioning based on an object data set of identified object classes and based on word embedding concepts, the object detection network configured to generalize to untrained object classes by the concept conditioning;
receiving an object search concept as a word input from which a word embedding is generated, the object search concept related to the particular image object depicted in the input digital image, the word embedding indicating a relationship between the object search concept and different word-based concepts; and
generating an output digital image from the input digital image and the word embedding, the output digital image depicting the scene and including one or more indications of object detection results that denote regions of the scene, including at least an indication of the particular image object corresponding to the object search concept.