From foot to head: Active face finding using deep Q-learning

In the existing work on active face detection and tracking, it is usually required that the face has to appear in the field-of-view. However, this may not be practical in some challenging scenarios. In this paper, we formulate the problem of active face finding as a Markov Decision Process and resort to the deep Q-learning to solve it in an end-to-end manner. Under the proposed framework, the agent is able to learn how to adjust the control parameters of a camera in order to find the face. Even if the captured image contains only some parts of the person, the PTZ camera can still adjust its pose until the face is found. Extensive experimental validations are performed to show the effectiveness of the developed system.

[1]  Reinaldo A. C. Bianchi,et al.  Automatic Selection of Object Recognition Methods Using Reinforcement Learning , 2010, Advances in Machine Learning I.

[2]  Katsunari Shibata,et al.  Acquisition of Context-Based Active Word Recognition by Q-Learning Using a Recurrent Neural Network , 2013, RiTA.

[3]  Di Guo,et al.  Weakly Paired Multimodal Fusion for Object Recognition , 2018, IEEE Transactions on Automation Science and Engineering.

[4]  Shane Legg,et al.  Human-level control through deep reinforcement learning , 2015, Nature.

[5]  Shengyong Chen,et al.  Active vision in robotic systems: A survey of recent developments , 2011, Int. J. Robotics Res..

[6]  Josef Kittler,et al.  Accurate face localisation for faces under active near-IR illumination , 2006, 7th International Conference on Automatic Face and Gesture Recognition (FGR06).

[7]  Andrew D. Bagdanov,et al.  Acquisition of high-resolution images through on-line saccade sequence planning , 2005 .

[8]  Lucas Paletta,et al.  Active object recognition by view integration and reinforcement learning , 2000, Robotics Auton. Syst..

[9]  Gamini Dissanayake,et al.  Active recognition and pose estimation of household objects in clutter , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[10]  Emanuele Menegatti,et al.  Cooperative tracking of moving objects and face detection with a dual camera sensor , 2010, 2010 IEEE International Conference on Robotics and Automation.

[11]  Thomas B. Moeslund,et al.  Real-time acquisition of high quality face sequences from an active pan-tilt-zoom camera , 2013, 2013 10th IEEE International Conference on Advanced Video and Signal Based Surveillance.

[12]  Fuchun Sun,et al.  Visual–Tactile Fusion for Object Recognition , 2017, IEEE Transactions on Automation Science and Engineering.

[13]  Javier R. Movellan,et al.  Deep Q-learning for Active Recognition of GERMS: Baseline performance on a standardized dataset for active learning , 2015, BMVC.

[14]  Gaurav S. Sukhatme,et al.  Active multi-view object recognition: A unifying view on online feature selection and view planning , 2016, Robotics Auton. Syst..

[15]  Nicholas Roy,et al.  Trajectory Optimization using Reinforcement Learning for Map Exploration , 2008, Int. J. Robotics Res..

[16]  Peter I. Corke,et al.  Towards Vision-Based Deep Reinforcement Learning for Robotic Motion Control , 2015, ICRA 2015.

[17]  Masakazu Matsugu,et al.  Face Tracking Active Vision System with Saccadic and Smooth Pursuit , 2006, 2006 IEEE International Conference on Robotics and Biomimetics.

[18]  John K. Tsotsos,et al.  A Computational Learning Theory of Active Object Recognition Under Uncertainty , 2012, International Journal of Computer Vision.

[19]  Joachim Denzler,et al.  A Framework for Actively Selecting Viewpoints in Object Recognition , 2009, Int. J. Pattern Recognit. Artif. Intell..

[20]  Dirk Kraft,et al.  Enhanced 3D face processing using an active vision system , 2015, 2014 International Conference on Computer Vision Theory and Applications (VISAPP).

[21]  Eric Sommerlade,et al.  Gaze directed camera control for face image acquisition , 2011, 2011 IEEE International Conference on Robotics and Automation.

[22]  Heinrich Niemann,et al.  Handling Camera Movement Constraints in Reinforcement Learning Based Active Object Recognition , 2006, DAGM-Symposium.

[23]  Svetlana Lazebnik,et al.  Active Object Localization with Deep Reinforcement Learning , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).