Plus One Robotics, Inc.

The University of Texas System

Abhijit Majumdar

Jonathan Lwowski

Berat Alper Erol

Divya Bhaskaran

Mohammad Jamshidi

Benjamin Factor

Patrick Benavidez

Arman Rezakhani

Training images can be synthesized in order to obtain enough data to train a model (e.g., a neural network) to recognize various classifications of a type of object. Images can be synthesized by blending images of objects labeled using those classifications into selected background images. To improve results, one or more operations are performed to determine whether the synthesized images can still be used as training data, such as by verifying one or more objects of interested represented in those images is not occluded, or at least satisfies a threshold level of acceptance. The training images can be used with real world images to train the model.

Systems and methods for creating training data

The present disclosure describes various embodiments of surveillance systems and methods. In one such embodiment, an exemplary surveillance system includes at least one video camera configured to capture video data of a surveilled area; and a computing device that stores a surveillance program. An exemplary surveillance program includes computer-executable instructions configured to: analyze the video data captured by the at least one video camera; identify objects that enter the surveilled area and log a time at which the objects entered the surveilled area; determine an object type for each object; track the identified objects to determine a period of time the objects have been present within the surveilled area; and generate and transmit an alert for each identified object that has been present within the surveilled area for a period of time that exceeds a predetermined time threshold.

Surveillance systems and methods for automatic real-time monitoring

Growing community of inventors

San Antonio, TX, United States of America

Abhijit Majumdar