FPGA-based implementation of CNN for Covid-19 identification

Despite advancements in treatment, the Coronavirus (Covid-19) remains a significant respiratory threat following the 2019 pandemic. Continuous efforts to enhance detection and diagnosis are crucial. Traditional methods, such as chest radiological imaging and computed tomography, are hindered by manual examination, leading to delays and potential errors. This study introduces a rapid and precise Covid-19 detection system utilizing the AlexNet Convolutional Neural Network (CNN) model on the DE1 SoC FPGA platform. The methodology encompasses ten stages: acquisition of CXR image datasets, dataset segmentation, installation of the Caffe framework, LMDB dataset formatting, generation of dataset image means, CNN model training, extraction of model weights, evaluation via confusion matrices, OpenCL environment setup, and PipeCNN compilation. The anticipated outcomes include reduced operation times and enhanced accuracy.