A low-cost FPGA-based SVM classifier for melanoma detection

Abstract

Support Vector Machines (SVMs) are common machine learning tools with accurate classification. Hardware implementation of SVM classifiers for real-time applications can improve their computing performance and reduce power consumption. This study aims to develop a real-time embedded classifier to be implemented on a low-cost handheld device dedicated for early detection of melanoma. Melanoma is the most dangerous form of skin cancer, which is responsible for the majority of skin cancer related deaths. Therefore, the proposed device would be very beneficial in the primary care. In this paper, a hardware design is proposed to implement a linear binary SVM classifier in an FPGA targeting online melanoma classification. A recent hybrid Zynq platform is used for the implementation of the proposed system designed using the latest High Level Synthesis design methodology. The implemented system demonstrates high performance, low hardware resources utilization and low power consumption that meet vital embedded systems constraints.