Artificial neural networks are effectively utilized for driver drowsiness detection based on video or image capturing systems. The detection of eye closure serves as a key step in detecting drowsy driving, and deep convolutional neural network (CNN)-based models are reported to achieve excellent prediction accuracy. However, such models have complex computational requirements making them impossible to be implemented in resource constrained edge devices. At least part of the model has to run in high performance remote systems. So, its successful operation depends heavily on reliability of the network connection. Spiking neural network (SNN) is a potential alternative for CNN especially for edge implementations. Its inherent discontinuous spike-train representation of information results in lesser storage and processing demands. In this paper, an SNN-based model is developed for eye closure detection and its performance is compared with the CNN-based model using metrics such as accuracy, latency, and number of floating point operations involved.