As is known to all, GPU (Graphical Processing Unit) has been shown to have processing capabilities superior to the CPU in scientific computing .Although modern CPUs may have several cores to allow small scale parallelization, GPU has thousands or millions of data parallel threads. In this project we tried to compare video processing performance between Intel OpenCV and CUDA, which is a new hardware and software architecture for issuing and managing computations on GPU.
CUDA Motion Bench provides assistance for users with basic knowledge of image processing and motional object detection. This software is a beta version.
The sample application can be downloaded here. Simply unzip the archive into a folder and run GPU_CPU_Compare.exe .We run antivirus, so it should be safe.
Ps:If you get the message “fail to load video”,Please install a video decoder.For example:ffshow.exe.
CUDA_Motion_Bench.zip (2.6 MB)
software_user_guide.pdf (144 KB)