Institut für Kognitionswissenschaft

Institute of Cognitive Science

Osnabrück University navigation and search

Deutsch

Submenu on the topic Research

Position indicator

You are here: Home > Research > Computer Vision > Research > Interactive 3D Modelling > Interactive Feature Growing

Main content

Top content

Interactive feature growing for accurate object detection in megapixel images

EPIC@ECCV 2016 - Demonstration Video and Supplementary Polts

Abstract Automatic object detection in megapixel images is quite inaccurate and a time and memory expensive task, even with feature detectors and descriptors like SIFT, SURF, ORB, and KAZE. In this paper we propose an interactive feature growing process, which draws on the efficiency of the users’ visual system. The performance of the visual system in search tasks is not affected by the pixel density, so the users’ gazes are used to boost feature extraction for object detection. Experimental tests of the interactive feature growing process show an increase of processing speed by 50% for object detection in 20 megapixel scenes at an object detection rate of 95%. Based on this method, we discuss the prospects of interactive features, possible use cases and further developments.

Demonstration video showing our new interactive feature growing baes on users’ gazes.

Result Polt

Figure 3Fig. 3 Average results over all twelve scene-object detection tasks. †first task of each subject is excluded with respect to the learning curve

Supplementary Polts

Supplementary Figure 1
Supplementary Fig. 1 Results scene-object detection tasks object 1 in scene 1.

Supplementary Figure 2 Supplementary Fig. 2 Results scene-object detection tasks object 2 in scene 1.

Supplementary Figure 3 Supplementary Fig. 3 Results scene-object detection tasks object 3 in scene 1.

 

Supplementary Figure 4 Supplementary Fig. 4 Results scene-object detection tasks object 4 in scene 1.

 

Supplementary Figure 5 Supplementary Fig. 5 Results scene-object detection tasks object 1 in scene 2.

 

Supplementary Figure 6 Supplementary Fig. 6 Results scene-object detection tasks object 3 in scene 2.

 

Supplementary Figure 7 Supplementary Fig. 7 Results scene-object detection tasks object 4 in scene 2.

 

Supplementary Figure 8 Supplementary Fig. 8 Results scene-object detection tasks object 5 in scene 3.

 

Supplementary Figure 9 Supplementary Fig. 9 Results scene-object detection tasks object 6 in scene 3.

 

Supplementary Figure 10 Supplementary Fig. 10 Results scene-object detection tasks object 7 in scene 4.

 

Supplementary Figure 11 Supplementary Fig. 11 Results scene-object detection tasks object 8 in scene 4.

 

Supplementary Figure 12 Supplementary Fig. 12 Results scene-object detection tasks object 9 in scene 4.

 

† first task of each subject is excluded with respect to the learning curve. For interactive number of subjects = 10, for automatic number of reruns = 10

Reference

J. Schöning, P. Faion & G. Heidemann.
Interactive Feature Growing for Accurate Object Detection in Megapixel Images.
Computer Vision — ECCV 2016 Workshops, 9913 : 546-556, ISBN: 978-3-319-46604-0, 2016. Springer Nature.
| PDF | DOI | URL | BibTeX

Footer