Evaluation of Camera Pose Estimation Using Human Head Pose Estimation

Abstract We introduce and evaluate a novel camera pose estimation framework that uses the human head as calibration object. The proposed method facilitates extrinsic from 2D input images (NIR and/or RGB), while merely relying on detected head, without need for depth information. approach is applicable to single cameras or multi-camera networks. Our implementation fine-tuned deep learning-based facial landmark detector estimate 3D by fitting model landmarks. work focuses an evaluation of real recordings synthetic renderings determine accuracy results their applicability. assess robustness our against different parameters, such varying relative positions, variations models, face occlusions (by masks, sun glasses, etc.), potential biases variance among humans. Based experimental results, we expect be effective numerous use cases including automotive attention monitoring, robotics, VR/AR other scenarios where ease handling outweighs accuracy.