Lightweight two-stream convolutional face detection
Danai Triantafyllidou, Paraskevi Nousi, Anastasios Tefas · 2017
Video capturing using Unmanned Aerial Vehicles provides cinematographers with impressive shots but requires very adept handling of both the drone and the camera. Deep Learning techniques can be utilized in this process to facilitate the video shooting process by allowing the drone to analyze its input and make intelligent decisions regarding its flight path. Fast and accurate on-board face detection for example can lead the drone towards capturing opportunistic shots, e.g., close ups of persons of importance. However, the constraints imposed by the drones' on-board processing power and memory prohibit the utilization of computationally expensive models. In this paper, we propose a lightweight two-stream fully Convolutional Neural Network for face detection, capable of detecting faces in various settings in real-time using the limited processing power Unmanned Aerial Vehicles possess.