TY - GEN
T1 - Diagnosing error in object detectors
AU - Hoiem, Derek
AU - Chodpathumwan, Yodsawalai
AU - Dai, Qieyun
N1 - Funding Information:
This work was supported by NSF awards IIS-1053768 and IIS-0904209, ONR MURI Grant N000141010934, and a research award from Google.
PY - 2012
Y1 - 2012
N2 - 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.
AB - 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.
UR - http://www.scopus.com/inward/record.url?scp=84867841321&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84867841321&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-33712-3_25
DO - 10.1007/978-3-642-33712-3_25
M3 - Conference contribution
AN - SCOPUS:84867841321
SN - 9783642337116
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 340
EP - 353
BT - Computer Vision, ECCV 2012 - 12th European Conference on Computer Vision, Proceedings
T2 - 12th European Conference on Computer Vision, ECCV 2012
Y2 - 7 October 2012 through 13 October 2012
ER -