Visual tracking with online Multiple Instance Learning

By Boris Babenko, Ming-Hsuan Yang, Serge J. Belongie. 2009

1,998 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

End-to-End Learning of Visual Representations From Uncurated Instructional Videos (2019, neutral): “Multiple instance learning methods have been employed in many weakly-supervised video understanding problems including: person recognition in movies using scripts [12, 50, 61], anomaly detection [70], weakly supervised action classification [47, 68] and localization [17, 22, 82], co-reference resolu”
Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Control (2018, methodology): “The ﬁrst is a cost function that uses our registration-based method, trained in a fully self-supervised fashion, and the second is with a cost function that uses off-the shelf tracking from OpenCV [48].”
Fully-Convolutional Siamese Networks for Object Tracking (2016, methodology): “ scenario has been to learn a model of the object’s appearance in an online fashion using examples extracted from the video itself [1]. This owes in large part to the demonstrated ability of MILTrack [2], Struck [3], TLD [4] and KCF [5]. However, a clear de- ciency of using data exclusively from the”
Robustness via Retrying: Closed-Loop Robotic Manipulation with Self-Supervised Learning (2018, methodology): “It is possible to use off-the-shelf trackers [30, 31, 32, 33] to mitigate this issue.”
A survey of appearance models in visual object tracking (2013, neutral): “…et al. 2005], histogram of oriented gradients [Collins et al. 2005; Liu and Yu 2007; Yu et al. 2008], Gabor features with Gabor wavelets [Nguyen and Smeulders 2004], and Haar-like features with Haar wavelets [Babenko et al. 2009]) can be used by FSSL in an independent or interleaving manner.”
Handling label noise in video classification via multiple instance learning (2011, methodology): “MIL has been successfully applied to computer vision tasks including face detection [13, 28], pose detection [2], image categorization [7], segmentation [25], and tracking [3].”
On classification with bags, groups and sets (2014, background): “Weakly annotated data is also a benefit in tracking [3].”
Robust Visual Tracking via Structured Multi-Task Sparse Learning (2012, methodology): “In the MIL tracker (Babenko et al. 2009), the multiple instance learning method is extended to an online setting for object”
Distribution fields for tracking (2012, methodology): “The tables show superiority of DF with respect to MIL and MKT in all 11 videos.”
Robust visual tracking via multi-task sparse learning (2012, methodology): “Other methods (IVT, L 1 , L ∞1 , OAB, MIL, and Frag) fail to track the object reliably.”
Tracking Revisited Using RGBD Camera: Unified Benchmark and Baselines (2013, methodology): “For example, MIL, CT, Struck, and TVD do not have an occlusion handling mechanism which naturally lead to high Type II error.”
Joint Monocular 3D Vehicle Detection and Tracking (2018, background): “Some generic object trackers are trained entirely online, starting from the first frame of a given video [16, 1, 21].”
Forward-Backward Error: Automatic Detection of Tracking Failures (2010, background): “proaches [7, 4, 1, 2] on 6 video sequences from [14].”
Real time robust L1 tracker using accelerated proximal gradient approach (2012, methodology): “Online multiple instance learning is used in [2] to achieve robustness to occlusions and other image corruptions.”
Minimum error bounded efficient ℓ1 tracker with occlusion detection (2011, methodology): “Our proposed tracker is compared with seven latest state-of-the-art trackers named Incremental Visual Tracking (IVT) [20], Multiple Instance Learning (MIL) [2], Visual Tracking Decomposition (VTD) [13], Generalized Kernel Tracking (GKT) [21], L1 tracker (L1) [18], Covariance Tensor Learning (CTL) [2”
Robust object tracking via sparsity-based collaborative model (2012, methodology): “In fact, several trackers including the FragTrack method [1], the MIL tracking algorithm [4], the ℓ 1 tracking method [19] and our tracker are developed to solve this problem.”
LaSOT: A High-Quality Benchmark for Large-Scale Single Object Tracking (2018, background): “We evaluate 35 algorithms on LaSOT to provide extensive baselines, comprising deep trackers ( e.g ., MDNet [42], TRACA [5], CFNet [50], SiamFC [4], StructSiam [59], DSiam [16], SINT [49] and VITAL [48]), correlation ﬁl-ter trackers with hand-crafted features ( e.g ., ECO HC [7], DSST [8], CN [11], C”
Part-based multiple-person tracking with partial occlusion handling (2012, background): “On the other hand, discriminative tracking approaches with online learning have flourished recently [1, 2, 11, 6, 10, 8].”
Learning to Track at 100 FPS with Deep Regression Networks (2016, background): “during test time) [15,3,36,19], with no offline training being performed.”
Online Object Tracking: A Benchmark (2013, methodology): “, the OAB [22], IVT [47], MIL [5], L1 [40], and TLD [31] algorithms, which have been commonly used for evaluation.”