Segformer to Assess Structural Health Using Synthesized Data from Blender

Tripathi, Neelotpal (2023) Segformer to Assess Structural Health Using Synthesized Data from Blender. Masters thesis, Indian Institute of Technology Hyderabad.

Text
CE21MTECH14017.pdf - Submitted Version
Restricted to Registered users only until 30 June 2028.
Download (11MB) | Request a copy

Abstract

Computer vision techniques have become increasingly important in the field of structural health monitoring. However, one of the main challenges in this area is the availability of high-quality datasets for training machine learning models. To address this challenge, we explore the use of Blender software to generate artificial datasets for structural health monitoring applications. Specifically, we generated images with custom textures using Blender and automated the camera positioning and rendering process using the Blender API. These images were then used to train and fine-tune the SegFormer neural network architecture for semantic segmentation tasks in component and damage detection. Our results demonstrate the effectiveness of our approach, achieving high accuracy in both tasks. This study highlights the potential of using computer vision and machine learning techniques for improving structural health monitoring, enabling faster and more accurate detection of potential issues in the infrastructure.

[error in script]

IITH Creators: