Abstract:Accurate identification of structural displacements is important for structural state assessment and performance evaluation. This paper proposes a real-time structural displacement detection model based on computer vision and deep learning. The model consists of three stages: identification, tracking, and displacement resolution. First, the displacement target is identified and tracked by the improved YOLO v7 algorithm and the improved DeepSORT algorithm. Then, the Euclidean distance method based on inverse perspective mapping (IPM-ED) is proposed for the analytical conversion of the displacement. Next, the accuracy and effectiveness of this displacement detection model are evaluated through four groups of bamboo axial compression tests. A comparative analysis is conducted between the IPM-ED displacement analysis method and the commonly used ED displacement analysis method. Finally, the robustness of this method is tested by using a cable breakage test of a cable dome structure as an application case. The research results demonstrate that the maximum average error of the four groups of bamboo displacement tests is only 3.10 mm, and the maximum relative error of peak displacement is only 6.54 mm. The RMSE basically stays around 3.5 mm. The maximum displacement error in the application case is only 4.91 mm, with a maximum MAPE of 4.94%. In addition, the error percentage under the IPM-ED algorithm is basically within 5%, while the error percentage of the ED algorithm is more than 10%. The method in this paper achieves efficient and intelligent identification of structural displacements in a non-contact manner. The proposed method is suitable for environments where the contact displacement sensor is easily affected by vibration, the site is complex and requires additional displacement sensor fixing equipment, the displacement sensor with super-high structure is unsafe to deploy, and the contact displacement sensor in narrow space is inconvenient to deploy, so it has broad application prospects.

A Deep Learning Augmented Vision-Based Method for Measuring Dynamic Displacements of Structures in Harsh Environments

Variational Mode Decomposition-Based Multirate Data-Fusion Framework for Estimating Structural Dynamic Displacement by Integrating Vision- and Acceleration-Based Measurements

Model-informed Deep Learning Strategy with Vision Measurement for Damage Identification of Truss Structures

Vision-based Structural Displacement Measurement: System Performance Evaluation and Influence Factor Analysis

Identification of Structural Dynamic Characteristics Based on Machine Vision Technology

A Vision-Based System for Dynamic Displacement Measurement of Long-Span Bridges: Algorithm and Verification

Image-Based Structural Dynamic Displacement Measurement Using Different Multi-Object Tracking Algorithms

Vision and Vibration Data Fusion-Based Structural Dynamic Displacement Measurement with Test Validation

Intelligent Noncontact Structural Displacement Detection Method Based on Computer Vision and Deep Learning

Multi-Point Displacement Monitoring of Bridges Using A Vision-Based Approach

Vision‐based Displacement Measurement Enhanced by Super‐resolution Using Generative Adversarial Networks

Three-dimensional Structural Displacement Measurement Using Monocular Vision and Deep Learning Based Pose Estimation

Vision-based Structural Displacement Measurement under Ambient-Light Changes Via Deep Learning and Digital Image Processing

Target-free 3D tiny structural vibration measurement based on deep learning and motion magnification

Computer vision based target-free 3D vibration displacement measurement of structures

Accelerometer-assisted Computer Vision Data Fusion Framework for Structural Dynamic Displacement Reconstruction

Robust and versatile vision-based dynamic displacement monitoring of natural feature targets in large-scale structures

A Multi-channel Input Framework for Structure Displacement Response Prediction Using Convolutional Neural Network

Accurate Measurement of Bridge Vibration Displacement Via Deep Convolutional Neural Network.

Measurement of Dynamic Characteristics on Structure using Non-marker Vision-based Displacement Measurement System

Automatic Displacement and Vibration Measurement in Laboratory Experiments with A Deep Learning Method