US 12,406,374 B2
Discriminative 3D shape modeling for few-shot instance segmentation
Anoop Cherian, Belmont, MA (US); Tim Marks, Newton, MA (US); and Alan Sullivan, Middleton, MA (US)
Assigned to Mitsubishi Electric Research Laboratories, Inc.
Filed by Mitsubishi Electric Research Laboratories, Inc., Cambridge, MA (US)
Filed on Feb. 25, 2022, as Appl. No. 17/652,533.
Claims priority of provisional application 63/268,398, filed on Feb. 23, 2022.
Prior Publication US 2023/0267614 A1, Aug. 24, 2023
Int. Cl. G06T 7/11 (2017.01); B25J 9/16 (2006.01); G06T 7/155 (2017.01); G06T 7/50 (2017.01); G06T 7/60 (2017.01)
CPC G06T 7/11 (2017.01) [B25J 9/1697 (2013.01); G06T 7/155 (2017.01); G06T 7/50 (2017.01); G06T 7/60 (2013.01); G06T 2207/10028 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20152 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An imaging controller for segmenting instances from depth images including objects to be manipulated by a robot comprising:
an input interface configured to receive a depth image that includes objects;
a memory configured to store instructions and a neural network trained to segment instances from the objects in the depth image; and
a processor, coupled with the memory, configured to perform the instructions to segment a pickable instance using the trained neural network, wherein steps of the instructions comprise:
selecting a tallest point in the depth image;
defining a region using a shape such that the region surrounds the tallest point;
sampling points in the region of the depth image;
computing depth-geodesics between the tallest point and the sampled points;
submitting the depth-geodesics to the neural network to segment the pickable instance among instances of the objects in the depth image; and
an output interface configured to output a geometrical feature of the pickable instance to a manipulator controller of the robot.