Classification and extraction method of hidden dangers along railway lines based on semantic segmentation network

International Journal of Remote Sensing, Journal Year: 2024, Volume and Issue: 45(24), P. 9480 - 9512

Published: Oct. 8, 2024

Foreign objects invading high-speed railway lines can cause danger. One existing solution is to use remote sensing images analyse the dangerous areas along line, thereby providing a certain amount of investigation time. Considering spatial and temporal resolution characteristics technologies in identifying floating reality rapid land changes, this paper identifies on ground where may be generated by using semantic segmentation techniques oriented remotely sensed imagery provides early warnings staff route. However, these regions that need analysed have different semantics scales. To address challenges, proposes Dual-branch Parallel Fusion Network (DPFNet) based Transformer, aimed at enhancing multi-class images. leverage global contextual information, we introduce Swin Transformer-based backbone network, which employs self-attention capture comprehensive scene context, facilitating better considering entire scene's context. For multi-scale features, propose one approach involves independent branching feature expression Multi-scale Feature Space Module (MFSFM). The former enriches while latter fuses features across levels diverse features. Experimental results demonstrate DPFNet effectively identify hidden danger area, fusion makes network more accurately segment risk area sizes, improving accuracy robustness, great significance formation 'prevention' as core safety operation.

Language: Английский

Neural Network-Based Bandit: A Medium Access Control for the IIoT Alarm Scenario

IEEE Open Journal of the Communications Society, Journal Year: 2024, Volume and Issue: 5, P. 7511 - 7524

Published: Jan. 1, 2024

Language: Английский