Object Tracking through Residual and Dense LSTMs

TL;DR

This paper enhances deep learning-based object trackers by integrating residual and dense LSTMs with skip connections, improving robustness and convergence in challenging scenarios like occlusions.

Contribution

It introduces residual and dense LSTM architectures with skip connections to improve the depth, convergence, and robustness of hybrid object trackers.

Findings

DenseLSTMs outperform Residual and regular LSTMs in tracking accuracy.

DenseLSTMs show higher resilience to occlusions and out-of-view objects.

Residual-based RNNs can enhance tracker robustness across different models.

Abstract

Visual object tracking task is constantly gaining importance in several fields of application as traffic monitoring, robotics, and surveillance, to name a few. Dealing with changes in the appearance of the tracked object is paramount to achieve high tracking accuracy, and is usually achieved by continually learning features. Recently, deep learning-based trackers based on LSTMs (Long Short-Term Memory) recurrent neural networks have emerged as a powerful alternative, bypassing the need to retrain the feature extraction in an online fashion. Inspired by the success of residual and dense networks in image recognition, we propose here to enhance the capabilities of hybrid trackers using residual and/or dense LSTMs. By introducing skip connections, it is possible to increase the depth of the architecture while ensuring a fast convergence. Experimental results on the Re3 tracker show that DenseLSTMs outperform Residual and regular LSTM, and offer a higher resilience to nuisances such as occlusions and out-of-view objects. Our case study supports the adoption of residual-based RNNs for enhancing the robustness of other trackers.