Learning Viewpoint Planning in Active Recognition on a Small Sampling Budget: A Kriging Approach

This paper focuses on viewpoint planning for 3D active object recognition. The objective is to design a planning policy into a Q-learning framework with a limited number of samples. Most existing stochastic techniques are therefore inapplicable. We propose to use Kriging and bayesian Optimization coupled with Q-learning to obtain a computationally-efficient viewpoint-planning design, under a restrictive sampling budget. Experimental results on a representative database, including a comparison with classical approaches, show promising results for this strategy.

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

[2]  Dana H. Ballard,et al.  Learning to perceive and act by trial and error , 1991, Machine Learning.

[3]  Peter Dayan,et al.  Q-learning , 1992, Machine Learning.

[4]  G. Matheron.  Principles of geostatistics , 1963 .

[5]  Nando de Freitas,et al.  A Bayesian exploration-exploitation approach for optimal online sensing and planning with a visually guided mobile robot , 2009, Auton. Robots.

[6]  Frank P. Ferrie,et al.  Autonomous recognition: driven by ambiguity , 1996, Proceedings CVPR IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[7]  Joachim Denzler,et al.  Integrated Viewpoint Fusion and Viewpoint Selection for Optimal Object Recognition , 2006, BMVC.

[8]  C. D. Perttunen,et al.  Lipschitzian optimization without the Lipschitz constant , 1993 .

[9]  Frank P. Ferrie,et al.  Viewpoint selection by navigation through entropy maps , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[10]  Donald R. Jones,et al.  Efficient Global Optimization of Expensive Black-Box Functions , 1998, J. Glob. Optim..

[11]  Joachim Denzler,et al.  Classifier Independent Viewpoint Selection for 3-D Object Recognition , 2000, DAGM-Symposium.

[12]  Michael James Sasena,et al.  Flexibility and efficiency enhancements for constrained global design optimization with kriging approximations. , 2002 .

[13]  Joachim Denzler,et al.  Information Theoretic Sensor Data Selection for Active Object Recognition and State Estimation , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[14]  Cordelia Schmid,et al.  A performance evaluation of local descriptors , 2005, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[15]  H. Roussel,et al.  Prediction from wrong models: the Kriging approach , 1996 .