Removing Mistracking of Multibody Motion Video Database Hopkins155

Abstract

Many mathematical techniques have been presented for classifying feature point trajectories over multibody motion video sequences into different motions, and most are applied to the Hopkins155 database for evaluating their performance. In this paper, we point out that Hopkins155 has problems and that it cannot necessarily evaluate the performance correctly. We create a new database by removing incorrect trajectories from Hopkins155. The basic principle of mistracking removal lies on the fact that correct trajectories all belong to parallel 2-D affine spaces in a high-dimensional space if all motions are translational and that parallel 2-D affine spaces are included in a 3-D affine space. Noting that if the image sequence is divided into short intervals, individual motions can be regarded as approximately translational in each interval, we detect incorrect trajectories by repeated plane fitting in the 3-D space using RANSAC. We point out why conventional outlier removal procedure does not work and demonstrate that out method allows us to tell in which frames incorrect trajectories occurred. The performance of multibody motion segmentation can be correctly evaluated using our database.

GORFLM: Globally Optimal Robust Fitting for Linear Model

“…In this experiment, we evaluated the performance of GORFLM on video motion segmentation with outliers. We used the traffic2 group from the Hopkins 155 data-set [43], corrected by removing mistrackings using the method in [44]. The traffic2 group consists of 31 sequences, each containing 2 motions.…”

T-Linkage: A Continuous Relaxation of J-Linkage for Multi-model Fitting

“…Surprsingly, it turnes out that in some cases the ratio is greater than one, meaning that the "second" Table 5: Motion segmentation: misclassification error (%) for video sequences with three motions closest model is actually closer than the ground-truth model. The presence of such equivocal points in the dataset might be related to the problems mentioned in a very recent paper [17] which we plan to consider in the forthcoming work.…”

Multiple Models Fitting as a Set Coverage Problem

“…We use the 51 real video sequences from the Hopkins 155 data-set [31], each containing two or three moving objects, with no outliers. Following [26], in order to deal with degenerate motions, we project the data onto an affine 4-d space where the rigidbody segmentation is translated in a 3-d plane fitting problem. Figure 4 reports some sample results, in particular three sequences belonging to Traffic 2 and Others 3 subsets, respectively, where ILP-RansaCov achieves sub-optimal segmentations.…”