Debugging Object Tracking Results by a Recommender System with Correction Propagation
Achieving error-free object tracking is almost impossible for state-of-the-art tracking algorithms in challenging scenarios such as tracking a large amount of cells over months in microscopy image sequences. Meanwhile, manually debugging (verifying and correcting) tracking results object-by-object and frame-by-frame in thousands of frames is too tedious. In this paper, we propose a novel scheme to debug automated object tracking results with humans in the loop. Tracking data that are highly erroneous are recommended to annotators based on their debugging histories. Since an error found by an annotator may have many analogous errors in the tracking data and the error can also affect its nearby data, we propose a correction propagation scheme to propagate corrections from all human annotators to unchecked data, which efficiently reduces human efforts and accelerates the convergence to high tracking accuracy. Our proposed approach is evaluated on three challenging datasets. The quantitative evaluation and comparison validate that the recommender system with correction propagation is effective and efficient to help humans debug tracking results.
M. Li and Z. Yin, "Debugging Object Tracking Results by a Recommender System with Correction Propagation," Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9009, pp. 214-228, Springer Verlag, Nov 2015.
The definitive version is available at http://dx.doi.org/10.1007/978-3-319-16631-5_16
Keywords and Phrases
Algorithms; Computer Vision; Errors; Image Processing; Recommender Systems; Tracking (Position); Large Amounts; Microscopy Images; Object Tracking; Quantitative Evaluation; State of the Art; Tracking Accuracy; Tracking Algorithm; Tracking Data; Program Debugging
International Standard Book Number (ISBN)
International Standard Serial Number (ISSN)
Article - Conference proceedings
© 2015 Springer Verlag, All rights reserved.