Research Direction:

AI in Computer Vision: 2D/3D scene understanding, saliency, object detection and recognition A multi-stage approach is proposed for salient object detection in natural images which

incorporates color and depth information. In the first stage, color and depth channels are

explored separately through objectness-based measures to detect potential regions containing salient objects. This procedure produces a list of bounding boxes which are further filtered and refined using statistical distributions. The retained candidates from both color and depth

channels are then combined using a voting system. The final stage consists of combining the

extracted candidates from color and depth channels using a voting system that produces a final map narrowing the location of the salient object. Publications:

• F. Audet, M.S. Allili, and A.-M. Cretu, "Salient Object Localisation in Images by

Combining Objectness Clues in the RGB-D Space", Karray F., Campilho A., Cheriet F.

(eds) Image Analysis and Recognition. ICIAR 2017, pp. 247-255, Lecture Notes in

Computer Science, vol. 10317. Springer.

This project investigates novel approaches for building identification in aerial images, that

combine classical segmentation algorithms, user guided training approaches and supervised

learning solutions.

Support: Publications: • J. Tremblay-Gosselin and A.-M. Cretu, "A Supervised Training and Learning Method for Building Identification in Remotely Sensed Images", IEEE Int. Symp. Robotic and Sensors Environments, Washington, US, 21-23 Oct. 2013, pp.73-78. • A.-M. Cretu, and P. Payeur, "Building Detection in Aerial Images Based on Watershed and Visual Attention Feature Descriptors", Canadian Conf. Computer and Robot Vision, pp. 265-272, Regina, 2013. • A.-M. Cretu, "Evolving Sensor System Environments with Visual Attention: An Experimental Exploration", Proc. IEEE Symp. Robotic and Sensors Environments, pp. 97-102, Magdeburg, Germany, 16-18 Nov. 2012 (invited talk). Drawing inspiration from the significantly superior performance of humans to extract and

interpret visual information, the proposed research transposes the visual attention

mechanisms into biologically-inspired computational systems to develop new techniques and computational resources capable to interpret complex images with large variations in their

content and characteristics. The project is dedicated to the exploration, implementation,

refinement and validation of a series of biologically-inspired computational models for

attention and object recognition dedicated to specific tasks, such as building detection in the context of geo-imaging. Collaborators/ Support:

Publications:

• A.-M. Cretu, and P. Payeur, "Visual Attention Model with Context Learning for Building Detection in Satellite Images", Int. Journal Smart Sensing and Intelligent Systems, vol. 5, no.4, pp. 742-766, Dec. 2012. • M. I. Sina, P. Payeur, and A.-M. Cretu, "Object Recognition on Satellite Images with Biologically-Inspired Computational Approaches", IEEE Int. Symp. Applied Computational Intelligence and Informatics, pp. 81-86, Timisoara, Romania, 24-26 May 2012. • M. I. Sina, A.-M. Cretu, and P. Payeur, “Biological Visual Attention Guided Automatic Image Segmentation with Application in Satellite Imaging”, Proc. IS&T/SPIE Electronic Imaging Conference, Human Vision and Electronic Imaging Track, vol. 8291, Burlingame, California, USA, Jan. 2012. The automated servicing of vehicles is becoming more and more a reality in today’s world. While certain operations,

such as car washing, require only a rough model of the

surface of a vehicle, other operations, such as changing of

a wheel or filling the gas tank, require correct localization of the different parts of the vehicle on which operations are to

be performed. The proposed image-based approach to

roughly localize vehicle parts over the surface of a vehicle

with a bounding box approach is based on a model of

human visual attention. The proposed method is

automatically adapted for different views of a vehicle and

obtains average localization rates for different vehicle parts

of over 95% for a dataset of 120 vehicles belonging to three categories, namely sedan, SUV and wagon and allows, with the addition of the

active contour models, for a more complete and accurate description of vehicle parts contours than other state-of-the-art solutions. Collaborators/ Support:

Publications:

• D. Nakhaeinia, P. Payeur, A. Chávez-Aragón, A.-M. Cretu, R. Laganière, and R. Macknojia, “Surface Following with an RGB-D Vision- Guided Robotic System for Vehicle Security Screening”, Int. Journal Smart Sensing and Intelligent Systems, vol. 9, no. 2, pp. 419-447, Jun. 2016. • A.-M. Cretu, P. Payeur, R. Laganière, "An Application of a Bio-Inspired Visual Attention Model Guided for the Localization of Vehicle Parts", Applied Soft Computing, vol.31. pp. 369-380, 2015. • R. Fareh, P. Payeur, D. Nakhaeinia, R. Macknojia, A. Chávez-Aragón, A.-M. Cretu, P. Laferrière, R. Laganière. R. Toledo, "An Integrated Vision-Guided Robotic System for Rapid Vehicle Inspection", IEEE Int. Systems Conference, SysCon, Ottawa, Mar. 2014 (Best Paper Award). • A.-M. Cretu, P. Payeur, "Image-Based Localization of Vehicle Parts Guided by Visual Attention", Proc. IEEE Int. Conf. Instrumentation and Measurement Technology, pp. 533-538, Graz, Austria, May 2012. The continuous rise in the amount of vehicles in circulation brings an increasing need for automatically and efficiently recognizing vehicle categories for multiple applications such as optimizing available parking spaces, balancing ferry

space, perceiving highway toll, planning infrastructure and managing traffic, or servicing vehicles. This project describes the design and implementation of a vehicle classification system using a set of images collected from 6 views. The proposed computational system combines human visual attention mechanisms to identify a set of salient discriminative features and a series of binary support vector machines to achieve fast automated classification. Collaborators/ Support:

Publications:

• A.-M. Cretu, and P. Payeur, "Biologically-Inspired Visual Attention Features for a Vehicle Classification Task", Int. Journal Smart Sensing and Intelligent Systems, vol. 4, no. 3, pp. 402-423, Sep. 2011. • A.-M. Cretu, P. Payeur, and R. Laganière, "Salient Features Based on Visual Attention for Multi-View Vehicle Classification", Proc. IEEE Int. Conf. Computational Intelligence for Measurement Systems and Applications, pp. 64-69, Ottawa, Canada, Sep. 2011. This project explores some aspects of intelligent sensing for

advanced robotic applications, with the main objective of

designing innovative approaches for automatic selection of

regions of observation for fixed and mobile sensors to collect

only relevant measurements without human guidance. The

proposed neural gas network solution selects regions of

interest for further sampling from a cloud of sparsely collected 3D measurements. The technique automatically determines

bounded areas where sensing is required at higher resolution to accurately map 3D surfaces. Therefore it provides significant benefits over brute force strategies as scanning time is reduced and the size of the dataset is kept manageable. Collaborators/ Support:

Publications:

• P. Payeur, P. Curtis, A.-M. Cretu, "Computational Methods for Selective Acquisition of Depth Measurements: an Experimental Evaluation", Int. Conf. Advanced Concepts for Intelligent Vision Systems, Poznan, Poland, J. Blanc-Talon et al. (Eds.) LNCS 8192, pp. 389-401, 2013. • P. Payeur, P. Curtis, A.-M. Cretu, "Computational Methods for Selective Acquisition of Depth Measurements in Machine Perception", IEEE Int. Conf. Systems, Man, and Cybernetics, Manchester, UK, pp.876-881, 2013. • A.-M. Cretu, P. Payeur, and E.M. Petriu, “Selective Range Data Acquisition Driven by Neural Gas Networks”, IEEE Trans. Instrumentation and Measurement, vol. 58, no. 6, pp. 2634-2642, 2009. • A.-M. Cretu, P. Payeur, and E.M. Petriu, “Neural Gas and Growing Neural Gas Networks for Selective 3D Sensing: A Comparative Study”, Sensors & Transducers, vol. 5, pp. 119-134, 2009. • A.-M. Cretu, P. Payeur, and E.M. Petriu, “Selective Tactile Data Acquisition on 3D Deformable Objects for Virtualized Reality Applications”, Proc. IEEE Int. Workshop Computational Intelligence in Virtual Environments, pp.14-19, Nashville, TN, USA, Apr. 2009 • A.-M. Cretu, E.M. Petriu, and P. Payeur, “Growing Neural Gas Networks for Selective 3D Scanning”, Proc. IEEE Int. Workshop on Robotic and Sensors Environments, pp. 108-113, Ottawa, Canada, Oct. 2008. • A.-M. Cretu, P. Payeur, and E.M. Petriu, “Selective Vision Sensing Based on Neural Gas Network”, Proc. IEEE Int. Conf. Instrumentation and Measurement Technology, pp. 478-483, Vancouver, Canada, May 2008 (Best Student Paper Award).