Efficient Visual Fault Detection for Freight Train via Neural Architecture Search with Data Volume Robustness

TL;DR

This paper introduces an efficient neural architecture search framework tailored for visual fault detection in freight trains, emphasizing scale-awareness and robustness to data volume, resulting in improved accuracy and reduced search costs.

Contribution

It proposes a novel NAS framework with a scale-aware search space and data volume robustness, enhancing detection performance and efficiency in freight train fault detection.

Findings

Achieves 46.8 and 47.9 mAP on Bottom View and Side View datasets.

Outperforms state-of-the-art methods in fault detection accuracy.

Reduces search costs linearly with decreased data volumes.

Abstract

Deep learning-based fault detection methods have achieved significant success. In visual fault detection of freight trains, there exists a large characteristic difference between inter-class components (scale variance) but intra-class on the contrary, which entails scale-awareness for detectors. Moreover, the design of task-specific networks heavily relies on human expertise. As a consequence, neural architecture search (NAS) that automates the model design process gains considerable attention because of its promising performance. However, NAS is computationally intensive due to the large search space and huge data volume. In this work, we propose an efficient NAS-based framework for visual fault detection of freight trains to search for the task-specific detection head with capacities of multi-scale representation. First, we design a scale-aware search space for discovering an effective receptive field in the head. Second, we explore the robustness of data volume to reduce search costs based on the specifically designed search space, and a novel sharing strategy is proposed to reduce memory and further improve search efficiency. Extensive experimental results demonstrate the effectiveness of our method with data volume robustness, which achieves 46.8 and 47.9 mAP on the Bottom View and Side View datasets, respectively. Our framework outperforms the state-of-the-art approaches and linearly decreases the search costs with reduced data volumes.