EAFD: An Edge Aware Lightweight Tiny Face Detector

Tiny Face detection, a crucial component of computer vision with many related applications, has seen significant advancements, particularly in the era of deep learning. However, existing Convolutional Neural Network (CNN) methods either rely on complex network architecture or bring in longer inference time. In this paper, we propose an efficient tiny face detector with limited resources and less inference time. Our method is based on an important observation that keeping or even enhancing edges is of great significance in tiny face detection. The introduction of several 1D and 2D filters streamlines our network architecture. Experiments on several real-time face databases demonstrate that our proposed face detector outperforms most state-in-art methods.