Reptile Meta-Tracking

Generic object tracking (GOT) is one of the main topics in computer vision for many years. The goal of GOT is to recognize and locate a specific object in the form of bounding box throughout a sequence of images. Moreover, GOT also requires algorithms to locate objects down to instances level. These requirements produce some unique challenges especially for deep learning based GOT algorithms that may easily become over-fitting if given a really small training dataset of the object during the online tracking process. To deal with this issue, we propose a novel Reptile meta-tracking algorithm, which adopts a first-order meta-learning technique so that during initialization, the visual tracker only requires few training examples and few steps of optimization to perform well. The proposed Reptile meta-tracker is evaluated on OTB2015 and VOT2018 tracking benchmark datasets, and outperforms several state-of-the-art trackers using one-pass evaluation.

[1]  Michael Felsberg,et al.  ECO: Efficient Convolution Operators for Tracking , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Silvio Savarese,et al.  Learning to Track at 100 FPS with Deep Regression Networks , 2016, ECCV.

[3]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[4]  Michael Felsberg,et al.  Convolutional Features for Correlation Filter Based Visual Tracking , 2015, 2015 IEEE International Conference on Computer Vision Workshop (ICCVW).

[5]  Alexander C. Berg,et al.  Meta-Tracker: Fast and Robust Online Adaptation for Visual Object Trackers , 2018, ECCV.

[6]  Bohyung Han,et al.  Learning Multi-domain Convolutional Neural Networks for Visual Tracking , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  Ming-Hsuan Yang,et al.  Object Tracking Benchmark , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[8]  Jiri Matas,et al.  Discriminative Correlation Filter Tracker with Channel and Spatial Reliability , 2016, International Journal of Computer Vision.

[9]  Sergey Levine,et al.  Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks , 2017, ICML.

[10]  J. Schulman,et al.  Reptile: a Scalable Metalearning Algorithm , 2018 .

[11]  Joshua B. Tenenbaum,et al.  Building machines that learn and think like people , 2016, Behavioral and Brain Sciences.

[12]  Jiri Matas,et al.  A Novel Performance Evaluation Methodology for Single-Target Trackers , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[13]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[14]  Xiaogang Wang,et al.  Visual Tracking with Fully Convolutional Networks , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[15]  Luca Bertinetto,et al.  Fully-Convolutional Siamese Networks for Object Tracking , 2016, ECCV Workshops.

[16]  Seunghoon Hong,et al.  Online Tracking by Learning Discriminative Saliency Map with Convolutional Neural Network , 2015, ICML.