English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Obstacles on the railway track leading to derailment accidents that cause significant damages to the railway in terms of killed and injuries over the years. Count of accident is increasing day by day due to its causes such as boulders on track, trees falling on the gauge, etc. Monitoring these events has been possible with humans working in railways. But when it comes to the real-time scenario, it turns to fatal work and requires more workers, particularly in a dangerous area. Also, this manual monitoring is not adequate to halt derailment accidents. In this perspective, railroad obstacle detection from aerial images has been growing as a trending research topic under artificial intelligence. Also, this mandates the assessment of familiar and latest deep neural network models such as CenterNet Hourglass, EfficientDet, Faster RCNN, SSD Mobile Net, SSD ResNet, and YOLO that detects the violator of accidents with the aid of our own developed Rail Obstacle Detection Dataset (RODD). These detectors were implemented on real-time aerial railway track images captured by Unmanned Aerial Vehicle (UAV) in India. Initially, the input images in the collected datasets were undergone to data preprocessing after that; the above mentioned deep neural models were trained individually. After that, the experiment is analyzed based on training, time, and performance metrics. At last, the results are visualized, evaluated, and compared; hence based on the performance, some effective deep neural network models have identified for detecting obstacles. The result shows that SSD Mobile Net and Faster RCNN can be used for railroad obstacle detection even in the different lighting conditions in railway with the accuracy of 96.75% and 84.75%, respectively.

A Comparative Assessment of Deep Neural Network Models for Detecting Obstacles in the Real Time Aerial Railway Track Images