Back to results
US 12,469,294 B2Granted
Computer vision-based system for controlling a garage door
OwnerChao Zhang, Zhongwei Cheng +3inventors
InventorsChao Zhang, Zhongwei Cheng, J Steven Skeoch, Lin Chen, Hongfei Ning
Life of this patent
Estimating the term…
Abstract
A method performed by an image capture device. The method includes determining whether to send a notification or determining whether to open a door based at least in part on at least one image captured by the image capture device.
Claims17
An image capture device for use with a door comprising at least one predefined feature, the image capture device comprising:
- at least one image sensor to capture images of the at least one predefined feature;
- a network interface to communicate with a computing device over a network;
- at least one processor;
- and
- memory comprising instructions to be executed by the at least one processor, the instructions when executed by the at least one processor causing the at least one processor to receive, from a user, an identification marking a location of the at least one predefined feature within at least one first image of the images captured by the at least one image sensor when the door is closed, train at least one machine learning process using the identification marking the location and the at least one first image as training data, use at least one second image to detect when the at least one predefined feature has moved, cause the trained at least one machine learning process to predict whether the door is in a closed state after detecting the at least one predefined feature has moved, and send a notification over the network interface to the computing device if the door is not predicted to be in the closed state.
2 more elements
A kit comprising:
- a marker to be affixed to a door;
- and
- a first image capture device to capture images of the marker, receive, from a user device, an identification labeling a portion of at least one first image of the images when the door is closed as including the marker, train at least one machine learning process using the identification labeling the portion and the at least one first image as training data, use at least one second image to detect when the marker has moved, cause the trained at least one machine learning process to predict whether the door is in a closed state after detecting the marker has moved, and send a notification to a computing device when the door is not predicted to be in the closed state.
A method comprising:
- capturing images, by an image capture device, of a region comprising a door;
- receiving, from a device operated by a user, an identification marking a location of at least one predefined feature within at least one first image of the images when the door is closed;
- training at least one machine learning process using the identification marking the location and the at least one first image as training data;
- using, by the image capture device, at least one second image to detect when the door has moved;
- causing, by the image capture device, at least one machine learning process to predict whether the door is in a closed state after detecting the at least one predefined feature has moved;
- and
- sending a notification, by the image capture device, to a computing device when the door is determined not to be in the closed state.
4 more elements
Dependencies are inferred from the claim text.