CPC G06F 16/5854 (2019.01) [G06F 3/0482 (2013.01); G06F 16/535 (2019.01); G06F 16/538 (2019.01); G06V 10/25 (2022.01); G06V 10/82 (2022.01); G06V 20/20 (2022.01)] | 20 Claims |
1. A method comprising:
receiving, by a processing device, image data captured by an image capture device that describes an environment scene including overlapping objects;
detecting, by the processing device, each object included in the overlapping objects by generating a bounding box for each object included in the overlapping objects using a region proposal network and a convolutional neural network;
generating, by the processing device, a mask for each object included in the overlapping objects by segmenting a foreground object in the bounding box for each object included in the overlapping objects;
receiving, by the processing device, device data describing a focus region of the image capture device;
determining, by the processing device, a default object of the overlapping objects by intersecting the mask for each object included in the overlapping objects with the focus region of the image capture device, the default object determined based on an overlap area of the mask for each object included in the overlapping objects with the focus region of the image capture device;
modifying, by the processing device, a visual attribute of one subject object that is included in the overlapping objects to expand or narrow results of an image search; and
generating, by the processing device, a search input for the image search, the search input includes the one subject object.
|