Learning Multi-domain Convolutional Neural Networks for Visual Tracking

TLDR

The authors introduce a visual tracking algorithm that leverages discriminatively trained CNN representations. The algorithm pretrains a CNN on a large video set to learn generic target features, then constructs a network with shared layers and domain‑specific binary classification branches that are iteratively trained; for a new sequence it combines the shared layers with an online‑updated classification layer and evaluates randomly sampled candidate windows around the previous target state. The algorithm achieves outstanding performance on standard tracking benchmarks.

Abstract

We propose a novel visual tracking algorithm based on the representations from a discriminatively trained Convolutional Neural Network (CNN). Our algorithm pretrains a CNN using a large set of videos with tracking ground-truths to obtain a generic target representation. Our network is composed of shared layers and multiple branches of domain-specific layers, where domains correspond to individual training sequences and each branch is responsible for binary classification to identify target in each domain. We train each domain in the network iteratively to obtain generic target representations in the shared layers. When tracking a target in a new sequence, we construct a new network by combining the shared layers in the pretrained CNN with a new binary classification layer, which is updated online. Online tracking is performed by evaluating the candidate windows randomly sampled around the previous target state. The proposed algorithm illustrates outstanding performance in existing tracking benchmarks.