Computer Vision and Artificial Intelligence Lab
In recent years wide area surveillance systems have gained significance for critical applications such as city security, traffic control, border security. These systems consist of multiple cameras with difference view angles or special sensors on the aerial vehicle patrolling a wide region. By stitching the camera images, a holistic view of the surveilled area is obtained. Hence it becomes possible to avoid criminal and terrorist activities, detect traffic violations on the patrolled region.
In this project we develop an embedded system for wide area surveillance by using together a wide field of view (FOV) camera and a scanning high resolution camera on the aerial platform. The developed system will be responsible for data acquisition from cameras, pre-processing (image stitching, ortho-rectification, etc.), data transfer to ground center, as well as exploitation of the imagery. For image exploitation purposes, object detection and tracking, anomaly/event detection and related surveillance and security applications will be implemented on embedded system under real-time constraint. For image and video analysis, image processing and deep learning techniques and architectures will be used. Hence the main goal is to design real-time and smart embedded system with advanced image and video analytics capabilities.
Publications:
Project Codes:
Datasets: