"This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by the authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be posted without the explicit permission of the copyright holder."
GPU-Based Complex-Background Segmentation Using Neural Networks
This page relates to the neural-network based foreground-background segmentation in videos containing complex background research conducted by our group at the Faculty of Technical Sciences in Novi Sad.
The page is intended to provide researchers with additional information and (eventually) the source code used in experiments described in our published papers, as well as those currently under review.
The sidebar lists various publications on the Background Modelling Neural Networks topic.
Below the publications you can find implementation of the algorithm designed to run on NVIDIA Graphics Processing Units (GPUs). The implementation is based on the OpenCV library and NVIDIA CUDA framework. The package currently contains no documentation but has a VS9.0 project file which will allow you to build it one the correct paths to CUDA an OpenCV libraries are set.
Several videos illustrating the segmentation results can be found below.
Please direct any inquiries to:
Several videos illustrating the foreground segmentation results can be found below.
Bridge surveillance sequence:
BNN output for the bridge sequence:
Final segmentation of the bridge sequence:
BNN output for water sequence:
Final segmentation output for the water sequence: