Fully-Convolutional Siamese Networks for Object Tracking

Abstract

The problem of arbitrary object tracking has traditionally been tackled by learning a model of the object's appearance exclusively online, using as sole training data the video itself. Despite the success of these methods, their online-only approach inherently limits the richness of the model they can learn. Recently, several attempts have been made to exploit the expressive power of deep convolutional networks. However, when the object to track is not known beforehand, it is necessary to perform Stochastic Gradient Descent online to adapt the weights of the network, severely compromising the speed of the system. In this paper we equip a basic tracking algorithm with a novel fully-convolutional Siamese network trained end-to-end on the ILSVRC15 dataset for object detection in video. Our tracker operates at frame-rates beyond real-time and, despite its extreme simplicity, achieves state-of-the-art performance in multiple benchmarks.

The first two authors contributed equally, and are listed in alphabetical order. Code available at http://www.robots.ox.ac.uk/~luca/siamese-fc.html

References in corpus (6)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• High-Speed Tracking with Kernelized Correlation Filters
• FlowNet: Learning Optical Flow with Convolutional Networks
• Transferring Rich Feature Hierarchies for Robust Visual Tracking
• Siamese Instance Search for Tracking
• First Step toward Model-Free, Anonymous Object Tracking with Recurrent Neural Networks

Cited by in corpus (13)

• Good Features to Correlate for Visual Tracking
• Deep Continuous Conditional Random Fields with Asymmetric Inter-object Constraints for Online Multi-object Tracking
• Large Margin Object Tracking with Circulant Feature Maps
• Learning Video Object Segmentation from Static Images
• Modeling Uncertainty with Hedged Instance Embedding
• Real-time visual tracking by deep reinforced decision making
• Siamese Natural Language Tracker: Tracking by Natural Language Descriptions with Siamese Trackers
• A Survey of Fish Tracking Techniques Based on Computer Vision
• Deep-LK for Efficient Adaptive Object Tracking
• DeepMix: Online Auto Data Augmentation for Robust Visual Object Tracking
• Deep Learning based Multi-Modal Sensing for Tracking and State Extraction of Small Quadcopters
• IG-TRACK: IOU Guided Siamese Networks for visual object tracking
• MobiFace: A Novel Dataset for Mobile Face Tracking in the Wild