In a typical face recognition pipeline, the task ofthe face detector is to localize the face region. However, the facedetector localizes regions that look like a face, irrespective of theliveliness of the face, which makes the entire system susceptible to presentation attacks. In this work, we try to reformulate thetask of the face detector to detect real faces, thus eliminatingthe threat of presentation attacks. While this task could bechallenging with visible spectrum images alone, we leverage themulti-channel information available from off the shelf devices(such as color, depth, and infrared channels) to design a multi-channel face detector. The proposed system can be used as alive-face detector obviating the need for a separate presentationattack detection module, making the system reliable in practicewithout any additional computational overhead. The main ideais to leverage a single-stage object detection framework, witha joint representation obtained from different channels for thePAD task. We have evaluated our approach in the multi-channelWMCA dataset containing a wide variety of attacks to show theeffectiveness of the proposed framework