Deep data association for online multi-class multi-object tracking
AssigneeXerox Corporation
InventorAdrien Gaidon
Life of this patent
Estimating the term…
Abstract
A system for applying video data to a neural network (NN) for online multi-class multi-object tracking includes a computer programed to perform an image classification method including the operations of receiving a video sequence; detecting candidate objects in each of a previous and a current video frame; transforming the previous and current video frames into a temporal difference input image; applying the temporal difference input image to a pre-trained neural network (NN) (or deep convolutional network) comprising an ordered sequence of layers; and based on a classification value received by the neural network, associating a pair of detected candidate objects in the previous and current frames as belonging to one of matching objects and different objects.
Claims19
A system for applying video data to a neural network (NN) for online multi-class multi-object tracking, the system comprising:
- a computer programed to perform a method for a classification of candidate object associations and including the operations of:
- receiving video frames from a video sequence;
- detecting candidate objects in each of a previous and a current video frame;
- transforming the previous and current video frames into a spatially-aligned temporal difference input image highlighting motion boundaries;
- applying the spatially-aligned temporal difference input image to a pre-trained neural network (NN) comprising an ordered sequence of layers;
- and
- based on a classification value received by the neural network, associating a pair of detected candidate objects in the previous and current frames as belonging to one of matching objects and different objects.
A non-transitory storage medium storing instructions readable and executable by a computer to perform a tracking method including the operations of:
- (a) detecting candidate objects in each of a previous and a current video frame;
- (b) transforming the previous and current video frames into a spatially-aligned temporal difference input image highlighting motion boundaries;
- (c) applying the temporal difference input image to a pre-trained neural network (NN) comprising an ordered sequence of layers;
- and
- (d) based on a classification value received by the neural network, associating a pair of detected candidate objects in the previous and current frames as belonging to one of matching objects and different objects.
Dependencies are inferred from the claim text.