Abstract:Deep neural networks, albeit their great success on feature learning in various computer vision tasks, are usually considered as impractical for online visual tracking, because they require very long training time and a large number of training samples. In this paper, we present an efficient and very robust tracking algorithm using a single convolutional neural network (CNN) for learning effective feature representations of the target object in a purely online manner. Our contributions are multifold. First, we introduce a novel truncated structural loss function that maintains as many training samples as possible and reduces the risk of tracking error accumulation. Second, we enhance the ordinary stochastic gradient descent approach in CNN training with a robust sample selection mechanism. The sampling mechanism randomly generates positive and negative samples from different temporal distributions, which are generated by taking the temporal relations and label noise into account. Finally, a lazy yet effective updating scheme is designed for CNN training. Equipped with this novel updating algorithm, the CNN model is robust to some long-existing difficulties in visual tracking, such as occlusion or incorrect detections, without loss of the effective adaption for significant appearance changes. In the experiment, our CNN tracker outperforms all compared state-of-the-art methods on two recently proposed benchmarks, which in total involve over 60 video sequences. The remarkable performance improvement over the existing trackers illustrates the superiority of the feature representations, which are learned purely online via the proposed deep learning framework.

Robust visual tracking via online semi-supervised co-boosting

Robust visual tracking via weighted incremental subspace learning

A new robust visual tracking algorithm based on Transfer Adaptive Boosting

Visual Tracking Using Online Semi-Supervised Learning

Scale and rotation invariant feature-based object tracking via modified on-line boosting

Robust Visual Tracking Via Transfer Learning

Robust Tracking with Weighted Online Structured Learning

Robust feature learning for online discriminative tracking without large-scale pre-training

Robust Online Visual Tracking Via A Temporal Ensemble Framework

Visual Tracking Jointly with Online and Offline Learning

Visual Tracking Via Weakly Supervised Learning from Multiple Imperfect Oracles

Tracking-assisted Weakly Supervised Online Visual Object Segmentation in Unconstrained Videos

A Bayesian Detect to Track System for Robust Visual Object Tracking and Semi-Supervised Model Learning

Robust and Real-Time Visual Tracking Based on Complementary Learners.

Discriminative and Robust Online Learning for Siamese Visual Tracking

Rapid visual tracking with modified on-line boosting and template matching

Robust Visual Tracking Via Multi-Graph Ranking

Feature-Based On-Line Object Tracking Combining Both Keypoints and Quasi-Keypoints Matching.

DeepTrack: Learning Discriminative Feature Representations Online for Robust Visual Tracking

Visual Tracking with Long-Short Term Based Correlation Filter

Robust Target Tracking by Online Random Forests and Superpixels