Object detection via a multi-region & semantic segmentation-aware CNN model

Spyros Gidaris (1, 2, 3) , Nikos Komodakis (4, 1, 2, 3, 5)

1 ENPC - École des Ponts ParisTech
2 LIGM - Laboratoire d'Informatique Gaspard-Monge
3 imagine [Marne-la-Vallée]
4 CVN - Centre de vision numérique
5 CSD-UOC - Computer Science Department [Crete]

We propose an object detection system that relies on a multi-region deep convolutional neural network (CNN) that also encodes semantic segmentation-aware features. The resulting CNN-based representation aims at capturing a diverse set of discriminative appearance factors and exhibits localization sensitivity that is essential for accurate object localization. We exploit the above properties of our recognition module by integrating it on an iterative localization mechanism that alternates between scoring a box proposal and refining its location with a deep CNN regression model. Thanks to the efficient use of our modules, we detect objects with very high localization accuracy. On the detection challenges of PASCAL VOC2007 and PASCAL VOC2012 we achieve mAP of 78.2% and 73.9% correspondingly, surpassing any other published work by a significant margin.

Mots clés: convolutional neural networks; deep learning; representation learning; Object detection; Biological system modeling; Semantics; Feature extraction; Computer architecture; Context; Shape

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]; Intelligence artificielle [cs.AI]

Spyros Gidaris, Nikos Komodakis. Object detection via a multi-region & semantic segmentation-aware CNN model. ICCV 2015, IEEE Computer Society, Dec 2015, Santiago, Chile. ⟨10.1109/ICCV.2015.135⟩. ⟨hal-01245664⟩