Incremental Learning of Linear Predictors for Fast Object Tracking
نویسنده
چکیده
This is a PhD thesis proposal. We concetrate on fast visual tracking methods in videos. We are especially interested in incremental learning of new object appearances. Current state of the art methods like background subtraction, kernel-based tracking or tracking by detection are shortly described. We use linear predictors for fast object tracking and an exhaustive description of the predictor learnig, tracking and incremental learning is given. We give also a detailed description of the most popular Lucas-Kanade tracking algorithm for comparison. Existing on-line learning algorithms are overviewed, too. In the final part of this text we summarize our current work and experiments and finaly we outline the doctoral thesis proposal, where we will focus on incremental learning of the linear predictor, automatic training examples selection and selection of good features to track.
منابع مشابه
Online Learning of Linear Predictors for Real-Time Tracking
Although fast and reliable, real-time template tracking using linear predictors requires a long training time. The lack of the ability to learn new templates online prevents their use in applications that require fast learning. This especially holds for applications where the scene is not known a priori and multiple templates have to be added online. So far, linear predictors had to be either l...
متن کاملFast Learnable Methods for Object Tracking
We propose a learning approach to tracking. The learning procedure explicitly minimizes the computational complexity of the tracking process subject to a user-defined probability of failure (loss-of-lock) and precision. In particular, we studied tracking methods estimating the object position by a learned linear mapping between intensities and motion. This mapping is called Learned Linear Predi...
متن کاملVisual Tracking with Online Incremental Deep Learning and Particle Filter
To solve the problem of tracking the trajectory of a moving object and learning a deep compact image representation in the complex environment, a novel robust incremental deep learning tracker is presented under the particle filter framework. The incremental deep classification neural network was composed of stacked denoising autoencoder, incremental feature learning and support vector machine ...
متن کاملLearning Efficient Linear Predictors for Motion Estimation
A novel object representation for tracking is proposed. The tracked object is represented as a constellation of spatially localised linear predictors which are learned on a single training image. In the learning stage, sets of pixels whose intensities allow for optimal least square predictions of the transformations are selected as a support of the linear predictor. The approach comprises three...
متن کاملFixed-point FPGA Implementation of a Kalman Filter for Range and Velocity Estimation of Moving Targets
Tracking filters are extensively used within object tracking systems in order to provide consecutive smooth estimations of position and velocity of the object with minimum error. Namely, Kalman filter and its numerous variants are widely known as simple yet effective linear tracking filters in many diverse applications. In this paper, an effective method is proposed for designing and implementa...
متن کامل