A unified light framework is designed to improve detection accuracy while supporting a real-time operation with a low resource requirement. We firstly design a novel lightweight backbone (RFDNet) to improve the accuracy and reduce computational cost. Then, we propose a multi region proposal network using multi-scale feature maps generated from RFDNet to improve the detection performance. Finally, we present multi level position-sensitive score maps and region of interest pooling to further improve accuracy with few redundant computations.

The proposed framework combines adjacency-graphs and kernel spectral clustering based graphs (KSC-graphs) according to a new definition named affinity nodes of multi-scale superpixels. These affinity nodes are selected based on a better affiliation of superpixels, namely subspace-preserving representation which is generated by sparse subspace clustering based on subspace pursuit. Then a KSC-graph is built via a novel kernel spectral clustering to explore the nonlinear relationships among these affinity nodes. Moreover, an adjacency-graph at each scale is constructed, which is further used to update the proposed KSC-graph at affinity nodes. The fusion graph is built across different scales, and it is partitioned to obtain final segmentation result.

Real-time vision based system of fault detection (RVBS-FD) for freight trains aims to complete routine maintenance tasks efficiently for ensuring railway security. Recently, the rapid development of deep learning techniques enables systems to provide a robust solution for the RVBS-FD of freight trains. We propose a CNN-based detector called Light FTI-FDet for the RVBS-FD of freight train. The results on five typical fault benchmarks indicate that our Light FTI-FDet achieves higher accuracy and fast speed with about 17% model size of the well-known Faster R-CNN detector, substantially outperforming state-of-the-art methods.

This paper proposes a novel unified framework for fault detection of the freight train images based on convolutional neural network under complex environment. Firstly, the multi region proposal networks with a set of prior bounding boxes are introduced to achieve high quality fault proposal generation. And then, we apply a linear non-maximum suppression method to retain the most suitable anchor while removing redundant boxes. Finally, a powerful multi-level region-of-interest pooling is proposed for proposal classification a nd accurate detection.

This paper proposes a novel hierarchical feature matching model for the typical faults detection, which is a big challenge in trouble of moving freight car detection system due to the constant color and complex background of images. The proposed model divides fault detection into two stages: image segmentation and parallel shape matching. In the process of segmentation, a fast adaptive Markov random field segmentation algorithm is presented based on MRF combined with the image Pyramid model and affinity propagation theory. In the process of shape matching, a novel shape descriptor named exact height function is introduced on the basis of height function description.