US 11,982,034 B2
Image quality detection for a camera assembly in a laundry appliance
Khalid Jamal Mashal, Louisville, KY (US)
Assigned to Haier US Appliance Solutions, Inc., Wilmington, DE (US)
Filed by Haier US Appliance Solutions, Inc., Wilmington, DE (US)
Filed on Feb. 3, 2021, as Appl. No. 17/166,228.
Prior Publication US 2022/0243377 A1, Aug. 4, 2022
Int. Cl. D06F 34/14 (2020.01); D06F 23/02 (2006.01); D06F 34/28 (2020.01); D06F 37/26 (2006.01); G06T 9/00 (2006.01); D06F 103/04 (2020.01); D06F 103/24 (2020.01); D06F 105/54 (2020.01); D06F 105/58 (2020.01); G06T 7/00 (2017.01); H04N 23/56 (2023.01)
CPC D06F 34/14 (2020.02) [D06F 23/02 (2013.01); D06F 34/28 (2020.02); D06F 37/266 (2013.01); G06T 9/002 (2013.01); D06F 2103/04 (2020.02); D06F 2103/24 (2020.02); D06F 2105/54 (2020.02); D06F 2105/58 (2020.02); G06T 7/0004 (2013.01); G06T 2200/24 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30164 (2013.01); G06T 2207/30168 (2013.01); H04N 23/56 (2023.01)] 11 Claims
OG exemplary drawing
 
1. A laundry appliance comprising:
a basket rotatably mounted within the cabinet and defining a chamber configured for receiving of a load of clothes;
a motor operably coupled to the basket for selectively rotating the basket;
a camera assembly mounted within the cabinet in view of the chamber; and
a controller operably coupled to the camera assembly, the controller being configured to:
determine that the basket is not rotating by monitoring operation of the motor or a basket speed sensor;
obtain an image of the chamber using the camera assembly;
determine that the basket is empty by analyzing the image;
analyze the image using an autoencoder neural network process programmed on the controller to determine an image quality, the analysis comprising performing an image reconstruction using the image of the chamber and the autoencoder neural network process and obtaining a reconstruction error based on a comparison of the reconstructed image to the image of the chamber;
determine that the image quality has dropped below a quality threshold based on the reconstruction error generated by the autoencoder neural network process; and
implement a responsive action in response to determining that the image quality has dropped below the quality threshold.