Relevant Motion Detection in Video
AssigneeComcast Cable Communications, LLC
InventorsRuichi Yu, Hongcheng Wang
Life of this patent
Estimating the term…
Abstract
Methods, systems, and/or apparatuses are described for detecting relevant motion of objects of interest (e.g., persons and vehicles) in surveillance videos. As described herein input data based on a plurality of captured images and/or video is received. The input data may then be pre-processed and used as an input into a convolution network that may, in some instances, have elements that perform both spatial-wise max pooling and temporal-wise max pooling. Based on The convolution network may be used to generate a plurality of prediction results of relevant motion of the objects of interest.
Claims1
A method comprising:
- generating, by a computing device and based on image data, data based on a plurality of images;
- determining, from the generated data and using one or more first convolutional layers, feature data indicative of one or more of:
- an appearance-related feature or a motion-related feature;
- determining, based on the feature data and using one or more second convolutional layers, predicted motion of one or more objects in the plurality of images;
- and
- determining, based on the predicted motion, that an object of the one or more objects has moved.
Dependencies are inferred from the claim text.