On-line Boosting and Vision

Visual tracking via weakly supervised learning from multiple imperfect oracles

“…For performance evaluation, we compare our approach against several representatives of the current state-of-the-art in visual tracking -the Fragments-based Tracker [5], the Online Boosting Tracker [3], the Semi-Supervised Online Boosting Tracker [8], the Beyond Semi-Supervised Tracker [9], the Online Multiple Instance Learning-based Tracker [10], and the SURF Tracker [16]. In the rest of our experiments, we refer to these six compared algorithms as FT, OBT, SSOBT, BSST, OMILT and ST respectively.…”

“…However, the method tends to have difficulties tracking objects that exhibit significant appearance changes. (2) Online Boosting Tracker [3]. The tracker uses online boosting method to obtain solutions.…”

“…(1) The set consists of a variety of complementary tracking approaches, such as gray histogram-based patch matching [5], motion consensus of local descriptors [16] and online classification using haar feature [3,8,9,10]. (2) The features used in all the six tracking methods can be extracted in a very efficient manner due to integral image data structure.…”

“…Challenges in visual tracking problems include non-rigid shape and appearance variations of the object, occlusions, illumination changes, cluttered scenes, etc. To solve them, most top-performing methods rely on online learning-based algorithms [1][2][3][4] to adaptively update target appearance. In these methods, visual tracking is formulated as an online binary classification problem and the target appearance is updated adaptively using the images tracked from the previous frames.…”

A Real-Time Human Detection System for Video

“…Okuma et al [18] employ a detector for hockey players and track them in a particle filter tracking framework. Breitenstein et al [19] utilize an off-line trained pedestrian detector [1] and online trained, instance-specific classifier via online boosting [20] for multi-person trackingby-detection. Another frequently used technique in video is background modeling, and most real-time systems [21], [22] depend on it for a fast speed.…”