CPC G06V 20/182 (2022.01) | 20 Claims |
1. A method, comprising:
receiving, by a device, tracking data identifying trajectories of objects;
annotating, by the device, the tracking data to identify object categories for the tracking data;
utilizing, by the device, an object detection model to identify the objects based on the object categories;
processing, by the device, the objects, with a multi-object tracking model, to determine object trajectories for the objects;
utilizing, by the device, planar homography to transform the object trajectories to overhead planar trajectories;
processing, by the device, the overhead planar trajectories, with a noise reduction model, to generate simplified object trajectories;
converting, by the device, a continuous space associated with the simplified object trajectories into a discrete space;
utilizing, by the device, a Hilbert curve with the discrete space to convert the discrete space to a lower dimensional space;
utilizing, by the device, the lower dimensional space and the discrete space to transform the object trajectories to binary feature vectors;
processing, by the device, the binary feature vectors, with a clustering model, to determine trajectory clusters;
training, by the device, a classification model, with the binary feature vectors and the trajectory clusters, to generate a trained classification model; and
performing, by the device, one or more actions based on the trained classification model.
|