Paper Info
Title
Diagnosing error in object detectors
Abstract
This paper shows how to analyze the influences of object characteristics on detection performance and the frequency and impact of different types of false positives. In particular, we examine effects of occlusion, size, aspect ratio, visibility of parts, viewpoint, localization error, and confusion with semantically similar objects, other labeled objects, and background. We analyze two classes of detectors: the Vedaldi et al. multiple kernel learning detector and different versions of the Felzenszwalb et al. detector. Our study shows that sensitivity to size, localization error, and confusion with similar objects are the most impactful forms of error. Our analysis also reveals that many different kinds of improvement are necessary to achieve large gains, making more detailed analysis essential for the progress of recognition research. By making our software and annotations available, we make it effortless for future researchers to perform similar analysis.
Year
DOI
Venue
2012
10.1007/978-3-642-33712-3_25
ECCV (3)
Keywords
Field
DocType
detailed analysis,similar analysis,diagnosing error,similar object,aspect ratio,object detector,different kind,different type,semantically similar object,detection performance,different version,localization error
Computer vision,Visibility,Confusion,Computer science,Multiple kernel learning,Software,Artificial intelligence,Detector,Machine learning,False positive paradox
Conference
Volume
ISSN
Citations 
7574
0302-9743
86
PageRank 
References 
Authors
15.09
22
3
Name
Order
Citations
PageRank
Derek Hoiem14998302.66
Yodsawalai Chodpathumwan210220.13
Qieyun Dai321719.85