Abstract:Visual tracking is a challenging task in computer vision. Recently correlation filter based trackers have gained much attention due to their high efficiency and impressive performance. Several methods have been developed to utilize deep features extracted from Convolutional Neural Networks (CNNs) for correlation filter tracking. Despite their success, most of these approaches suffer from low tracking speed due to high computation burden of the deep models. In this paper, we propose a CNN based real-time deep tracker for robust visual tracking. We first exploit hierarchical deep features as target representation, which can better distinguish the target from the backgrounds in the presence of complex situations. Before applied to learn correlation filters, the channel number of hierarchical features is reduced by the principal component analysis (PCA). This method can decrease both feature redundancy and computation. Then we construct a reliable map with features from the last convolutional layer as they encode the most semantic information. The reliable map is used to constraint the searching area where the target most likely exist. To further handle long-term tracking, we improve the model discrimination capability by introducing the original template into the current correlation filter model. As the target movement is usually small between two adjacent frames, we reuse the deep features from location step to update current model. The features are reused by making a circular shift with position displacement of the target, which increases the tracking speed significantly. Extensive experimental results on large benchmarks show that our proposed tracker can perform at real-time and achieves the state-of-the-art performance.

Deep People Counting with Faster R-CNN and Correlation Tracking.

Real Time Crowd Counting with Human Detection and Human Tracking.

Deep People Counting in Extremely Dense Crowds

Deep Correlation Filter based Real-Time Tracker

A Reliable People Counting System Via Multiple Cameras.

Counting people in crowded scenes by video analyzing

Depth-Assisted Face Detection and Association for People Counting.

Crowd Counting Using Deep Recurrent Spatial-Aware Network.

A People Counting Method Based on Head Detection and Tracking

Auxiliary Learning for Crowd Counting Via Count-Net

Deep Spatial Regression Model for Image Crowd Counting

Spatial-related and Scale-aware Network for Crowd Counting.

Density-Attentive Head Detector for Crowd Counting

Head-Body Correlation for Robust Crowd Human Detection

Real-time accurate crowd counting based on RGB-D information

A Real-Time Deep Network for Crowd Counting

Correlation-attention guided regression network for efficient crowd counting

Deep Convolution Network for Dense Crowd Counting.

Tracking People in Rgbd Videos Using Deep Learning and Motion Clues

A People Counting System Based on Face Detection and Tracking in a Video

Research on Local Counting and Object Detection of Multiscale Crowds in Video Based on Time-Frequency Analysis