Multi-block SSD based on small object detection for UAV railway scene surveillance

A method of multi-block Single Shot Multi Box Detector(SSD)based on small object detection is proposed to the railway scene of unmanned aerial vehicle surveillance.To address the limitation of small object detection,a multi-block SSD mechanism,which consists of three steps,is designed.First,the original input images are segmented into several overlapped patches.Second,each patch is separately fed into an SSD to detect the objects.Third,the patches are merged together through two stages.In the first stage,the truncated object of the sub-layer detection result is spliced.In the second stage,a sub-layer suppression and filtering algorithm applying the concept of non-maximum suppression is utilized to remove the overlapped boxes of sub-layers.The boxes that are not detected in the main-layer are retained.In addition,no sufficient labeled training samples of railway circumstance are available,thereby hindering the deployment of SSD.A two-stage training strategy leveraging to transfer learning is adopted to solve this issue.The deep learning model is preliminarily trained using labeled data of numerous auxiliaries,and then it is refined using only a few samples of railway scene.A railway spot in China,which is easily damaged by landslides,is investigated as a case study.Experimental results show that the proposed multi-block SSD method produces an overall accuracy of 96.6%and obtains an improvement of up to 9.2%compared with the traditional SSD.

A template-based knowledge reuse method for generating multitype 3D railway scenes

Constructing 3D railway scenes helps improve the railway information construction ability and the comprehensive railway management level.However,the existing modeling methods have disadvantages such as low flexibility in scene configuration,weak extensibility of scene objects and poor reusability of modeling knowledge.To enable multitype 3D railway scenes to be automatically constructed,a template-based knowledge reuse method was proposed in this paper.First,based on parsing of modeling operation characteristics and modeling knowledge expression,a modeling knowledge template was designed to store modeling knowledge and modeling operations in a parameterized way.The contents of this template were described in terms of semantic information,geometric information,topologic information and reference system information,and the method for instantiating the modeling knowledge template was proposed.Second,a scene configuration template was designed to organize the instantiated modeling knowledge files and scene objects.By the use of a flexible tree structure,the scene configuration template enables extensibility of scene objects.Finally,the two types of templates were parsed in terms of scene organization,linear referencing information,geometric and topologic relationships,and semantic descriptions.In this way,different types of 3D railway scenes were generated automatically.After a prototype system was developed,experiments on automatic modeling of two 3D railway scenes were carried out to verify the template-based knowledge reuse method.The experimental results show that the proposed method can be used to generate multitype 3D railway scenes automatically.With this method,model combination can be configured,model types can be extended and modeling knowledge can be reused.