Abstract
Image classifiers often rely on convolutional neural networks (CNN) for theirtasks, which are inherently more heavyweight than multilayer perceptrons(MLPs), which can be problematic in real-time applications. Additionally, manyimage classification models work on both RGB and grayscale datasets.Classifiers that operate solely on grayscale images are much less common.Grayscale image classification has diverse applications, including but notlimited to medical image classification and synthetic aperture radar (SAR)automatic target recognition (ATR). Thus, we present a novel grayscale (singlechannel) image classification approach using a vectorized view of images. Weexploit the lightweightness of MLPs by viewing images as a vector and reducingour problem setting to the grayscale image classification setting. We find thatusing a single graph convolutional layer batch-wise increases accuracy andreduces variance in the performance of our model. Moreover, we develop acustomized accelerator on FPGA for the proposed model with severaloptimizations to improve its performance. Our experimental results on benchmarkgrayscale image datasets demonstrate the effectiveness of the proposed model,achieving vastly lower latency (up to 16$\times$ less) and competitive orleading performance compared to other state-of-the-art image classificationmodels on various domain-specific grayscale image classification datasets.