In this paper, we present an effective method for addressing the face anti-spoofing (FAS) problem. Our proposed method is based on the simple concept of end-to-end binary supervision, enhanced with several improvements to boost model performance. First, features extracted from a CNN model are refined using an upgraded version of the spatial/channel-wise attention module. Second, we introduce auxiliary branches to supervise depth map estimation accuracy at two different levels, aiding in the effective capture of depth-related information and providing the model with additional cues to discriminate between real and spoofed faces. We evaluated our proposed method on various public datasets using both intra-testing and cross-testing scenarios. Experimental results demonstrate the effectiveness of our approach, outperforming both referenced methods and current state-of-the-art techniques.