GPU-Accelerated Foreground Segmentation and Labeling for Real-Time Video Surveillance

Song, Wei; Tian, Yifei; Fong, Simon; Cho, Kyungeun; Wang, Wei; Zhang, Weiqiang

doi:10.3390/su8100916

Cited by 12 publications

(7 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They improve the randomized expansion and updating the speed of their method by applying GPU accelerations. Song et al [26] proposed a parallel-connected component labeling method to segment foregrounds by using pixelwise color histograms in GPUs. Foreground segmentation results will be clustered to obtain separate different foreground objects.…”

Section: Related Workmentioning

confidence: 99%

A Content-Adaptive Resizing Framework for Boosting Computation Speed of Background Modeling Methods

Huang

Liao

et al. 2022

IEEE Trans. Syst. Man Cybern, Syst.

View full text Add to dashboard Cite

Recently, most background modeling (BM) methods claim to achieve real-time efficiency for low-resolution and standard-definition surveillance videos. With the increasing resolutions of surveillance cameras, full high-definition (full HD) surveillance videos have become the main trend and thus processing high-resolution videos becomes a novel issue in intelligent video surveillance. In this article, we propose a novel contentadaptive resizing framework (CARF) to boost the computation speed of BM methods in high-resolution surveillance videos. For each frame, we apply superpixels to separate the content of the frame to homogeneous and boundary sets. Two novel downsampling and upsampling layers based on the homogeneous and boundary sets are proposed. The front one downsamples highresolution frames to low-resolution frames for obtaining efficient foreground segmentation results based on BM methods. The later one upsamples the low-resolution foreground segmentation results to the original resolution frames based on the superpixels. By simultaneously coupling both layers, experimental results show that the proposed method can achieve better quantitative and qualitative results compared with state-of-the-art methods. Moreover, the computation speed of the proposed method without GPU accelerations is also significantly faster than that of the state-of-the-art methods. The source code is available at https://github.com/nchucvml/CARF. Index Terms-Background modeling (BM), frame downsampling, frame resizing, frame upsampling, superpixels. I. INTRODUCTION B ACKGROUND modeling (BM) methods are shown to be effective and efficient for foreground segmentation in intelligent video surveillance (IVS) [1]. It serves as one Manuscript

show abstract

Section: Related Workmentioning

confidence: 99%

A Content-Adaptive Resizing Framework for Boosting Computation Speed of Background Modeling Methods

Huang

Liao

et al. 2022

IEEE Trans. Syst. Man Cybern, Syst.

View full text Add to dashboard Cite

show abstract

“…With parallel computation, the time required decreased by more than 15 times, thereby achieving a fast clustering approach. To accelerate the labeling process, we employed a GPU-based CCL algorithm to cluster foreground areas in real-time surveillance videos [30]. Differing from image clustering processes, 3D LiDAR point clouds are dispersed and without structural and connected relationships among neighboring points.…”

Section: Related Workmentioning

confidence: 99%

A Fast Spatial Clustering Method for Sparse LiDAR Point Clouds Using GPU Programming

Tian

Song

Chen³

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

Fast and accurate obstacle detection is essential for accurate perception of mobile vehicles’ environment. Because point clouds sensed by light detection and ranging (LiDAR) sensors are sparse and unstructured, traditional obstacle clustering on raw point clouds are inaccurate and time consuming. Thus, to achieve fast obstacle clustering in an unknown terrain, this paper proposes an elevation-reference connected component labeling (ER-CCL) algorithm using graphic processing unit (GPU) programing. LiDAR points are first projected onto a rasterized x–z plane so that sparse points are mapped into a series of regularly arranged small cells. Based on the height distribution of the LiDAR point, the ground cells are filtered out and a flag map is generated. Next, the ER-CCL algorithm is implemented on the label map generated from the flag map to mark individual clusters with unique labels. Finally, obstacle labeling results are inverse transformed from the x–z plane to 3D points to provide clustering results. For real-time 3D point cloud clustering, ER-CCL is accelerated by running it in parallel with the aid of GPU programming technology.

show abstract

“…Researches on sustainability using GPUs in large computing systems [29,30] focus primarily on high performance computing (HPC). In general, software acceleration can replace hardware replacement by using parallel processing of existing GPUs in existing computing systems or personal computers.…”

Section: Related Workmentioning

confidence: 99%

Hardware-Based Adaptive Terrain Mesh Using Temporal Coherence for Real-Time Landscape Visualization

Lee¹,

Shin²

2019

Sustainability

View full text Add to dashboard Cite

In general, changes in society or the environment are expected depending on changes in terrain. The faster and more accurately these terrain changes can be observed, the faster and more accurately predictions can be made. Recently, three-dimensional (3D) terrain visualization programs, such as flight simulation, allow for interaction with various datasets to predict ecosystem influences in real time. Elaborate terrain data require a very large capacity. To render these large terrain data, the computing power of graphics devices cannot always satisfy the real-time conditions. Consequently, a large number of graphics devices in computing systems need to be replaced on a periodic basis. As the industry evolves, the replacement cycle of graphic devices shortens. To solve this problem, we present a novel acceleration approach for generating an adaptive terrain mesh using temporal coherence. By using our method, it is possible to prevent artifacts such as frame drop or screen flickering due to lack of computing power of the GPU in a specific viewing condition. Instead of generating the new terrain mesh on every frame, our method reuses the detail level of terrain mesh that was used in a previous frame. Therefore, it can maintain the frame coherency and improve the rendering speed. This allows the proposed method to more quickly provide more detailed information about the terrain to predict environmental changes more accurately on existing equipment. Thus, the proposed method can reduce the need to frequently replace GPUs. The proposed method can guarantee sufficient performance even with a resilient graphic device and can effectively slow down the replacement period of existing equipment.

show abstract

GPU-Accelerated Foreground Segmentation and Labeling for Real-Time Video Surveillance

Cited by 12 publications

References 41 publications

A Content-Adaptive Resizing Framework for Boosting Computation Speed of Background Modeling Methods

A Content-Adaptive Resizing Framework for Boosting Computation Speed of Background Modeling Methods

A Fast Spatial Clustering Method for Sparse LiDAR Point Clouds Using GPU Programming

Hardware-Based Adaptive Terrain Mesh Using Temporal Coherence for Real-Time Landscape Visualization

Contact Info

Product

Resources

About