Object detection with incomplete supervision

16 marzo 2015
16 marzo 2015

Time: 11:00
Location: Meeting Room Ofek, Polo Scientifico e Tecnologico "Fabio Ferrari" (Building Povo 1), via Sommarive 5 - Povo, Trento

Speaker

  • Jacob Verbeek
    INRIA Rhone-Alpes, Grenoble, France

Abstract
Object category localization is a challenging problem in computer vision. Standard supervised training requires bounding box annotations of object instances. This time-consuming annotation process is sidestepped in weakly supervised learning. In this case, the supervised information is restricted to binary labels that indicate the absence/presence of object instances in the image, without their locations. We follow a multiple-instance learning approach that iteratively trains the detector and infers the object locations in the positive training images. Our main contribution is a multi-fold multiple instance learning procedure, which prevents training from prematurely locking onto erroneous object locations. This procedure is particularly important when using high-dimensional representations, such as Fisher vectors and convolutional neural network features. We also propose a window refinement method, which improves the localization accuracy by incorporating an objectness prior. We present a detailed experimental evaluation using the VOC 2007 dataset, which verifies the effectiveness of our approach.

About the Speaker
Jakob Verbeek received a PhD degree in computer science in 2004 from the University of Amsterdam, The Netherlands. After being a post- doctoral researcher at the University of Amsterdam and at INRIA Rhone-Alpes, he has been a full-time researcher at INRIA, Grenoble, France, since 2007. 
His research interests include machine learning and computer vision, with special interest in applications of statistical models in computer vision.

Contact Person Regarding this Talk: Nicu Sebe, niculae.sebe [at] unitn.it