How to improve the HOG detector in the UAV context

Paul Blondel (1) , Alex Potelle (1) , Claude Pegard (1) , Rogelio Lozano (2)

1 MIS - Modélisation, Information et Systèmes - UR UPJV 4290
2 Heudiasyc - Heuristique et Diagnostic des Systèmes Complexes [Compiègne]

The well known HOG (Histogram of Oriented Gradients) of Dalal and Triggs is commonly used for pedestrian detection from 2d moving embedded cameras (driving assistance) or static cameras (video surveillance). In this paper we show how to use and improve the HOG detector in the UAV context. In order to increase the elevation angular robustness we propose to use a more appropriate training dataset and sliding windows. We show results on synthetic images.

Mots clés: supervised training; HOG; UAV; human detection; computer vision

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]; Apprentissage [cs.LG]

Paul Blondel, Alex Potelle, Claude Pegard, Rogelio Lozano. How to improve the HOG detector in the UAV context. 2nd IFAC Workshop on Research, Education and Development of Unmanned Aerial Systems (RED UAS 2013), Nov 2013, Compiègne, France. pp.46-51. ⟨hal-01270654⟩