Mean LBP and modified fuzzy C-means weighted hybrid feature for illumination invariant mean-shift tracking

Phadke, Gargi; Velmurugan, Rajbabu

doi:10.1007/s11760-016-1008-0

Cited by 12 publications

(1 citation statement)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Off-the-shelf versions of MeanShift tracking rely on a histogram calculated from the original frame throughout the whole scene. This does not work well if the illumination in the scene changes, since the algorithm cannot make fast updates to the histogram [30,84,58,59]. MeanShift++ can adapt to changing color distributions by finding and adding neighboring grids of points to the histogram in linear time, making it more robust to gradual changes in lighting, color, and other artifacts of the data.…”

Section: Object Trackingmentioning

confidence: 99%

MeanShift++: Extremely Fast Mode-Seeking With Applications to Segmentation and Object Tracking

Jang¹,

Jiang²

2021

Preprint

View full text Add to dashboard Cite

MeanShift is a popular mode-seeking clustering algorithm used in a wide range of applications in machine learning. However, it is known to be prohibitively slow, with quadratic runtime per iteration. We propose MeanShift++, an extremely fast mode-seeking algorithm based on Mean-Shift that uses a grid-based approach to speed up the mean shift step, replacing the computationally expensive neighbors search with a density-weighted mean of adjacent grid cells. In addition, we show that this grid-based technique for density estimation comes with theoretical guarantees. The runtime is linear in the number of points and exponential in dimension, which makes MeanShift++ ideal on lowdimensional applications such as image segmentation and object tracking. We provide extensive experimental analysis showing that MeanShift++ can be more than 10,000x faster than MeanShift with competitive clustering results on benchmark datasets and nearly identical image segmentations as MeanShift. Finally, we show promising results for object tracking.

show abstract

Section: Object Trackingmentioning

confidence: 99%