(RP18) Real-Time Fire Detection Using CUDA
Performance Analysis and Optimization
TimeTuesday, June 18th8:30am - 10am
LocationSubstanz 1, 2
DescriptionIn this research, a high-resolution real-time fire detection system was implemented. NVIDIA CUDA framework was used to parallelize a serial version that was implemented using OpenCV. The algorithm relies on color thresholding with other de-noising image processing techniques applied to track the fire. Both implementations of the fire detection algorithm were compared, and the reported results show that the parallel implementation achieved a 60% speedup over the serial version.