This study uses a novel directional lighting approach to produce a computationally efficient five-channel Visual Geometry Group-16 (VGG-16) convolutional neural network (CNN) model for concrete crack detection and classification in low-light environments. The first convolutional layer of the proposed model copies the weights for the first three channels from the pre-trained model. In contrast, the additional two channels are set to the average of the existing weights along the channel. The model employs transfer learning and fine-tuning approaches to enhance accuracy and efficiency. It utilizes variations in patterned lighting to produce five channels. Each channel represents a grayscale version of the images captured using directed lighting in the right, below, left, above, and diffused directions, respectively. The model is evaluated on concrete crack samples with crack widths ranging from 0.07 mm to 0.3 mm. The modified five-channel VGG-16 model outperformed the traditional three-channel model, showing improvements ranging from 6.5 to 11.7 percent in true positive rate, false positive rate, precision, F1 score, accuracy, and Matthew’s correlation coefficient. These performance improvements are achieved with no significant change in evaluation time. This study provides useful information for constructing custom CNN models for civil engineering problems. Furthermore, it introduces a novel technique to identify cracks in concrete buildings using directed illumination in low-light conditions.
External lighting is required for autonomous inspections of concrete structures in low-light environments; however, previous studies have only considered uniformly diffused lighting to illuminate images. This study proposes a novel algorithm that utilises angled and directional lighting to obtain pixel-level segmentation of concrete cracks. The method applies a concrete crack detection algorithm to separate images, each illuminated with lighting from a different direction. Using a bitwise OR operation, the findings from all images are combined; the resulting image highlights the extremities of any present cracks in all lighting directions. When tested on a dataset of cracks ranging in widths from 0.07 mm to 0.3 mm, the algorithm obtained recall, precision and F1 score results of 77%, 84% and 92%, respectively. The algorithm was able to correctly segment cracks that were deemed too thin for similar diffused lighting segmentation methods found in literature. The proposed directional lighting algorithm has the potential to improve concrete inspections in low-light environments.
This project aim to robotically deploy vibrating wireless strain gauges (VWSG) into small scale steel fibre reinforced concrete (SFRC) tunnel segment making it smart. The VWSGs connected to an autonomous wireless node can establish ad-hoc modular networks with other smart segments and the segment properties can be tracked through their whole lifecycle. The main objective pursued are: (i) the design, the implementation, and the performance assessment of the robotic process of installing the sensors; (ii) the design, the fabrication and the mechanical testing of smart segment under cyclical loadings.
Existing automated concrete inspection methods are intractable: capturing images under ambient conditions which can vary substantially. Furthermore, an opportunity may have been overlooked: utilizing illumination techniques to enhance defect contrast during imaging which may improve automatic defect detection accuracy. In this work, we present a robotic-mountable lighting apparatus that implements contrast enhancing illumination techniques in an automated package in order to improve crack detection and classification in concrete. Geometrical lighting techniques; directional and angled, were tested on three cracked concrete slab samples. Results from blind/reference less image spatial quality evaluation (BRISQUE) show that both directional and varied angled lighting influence the quality in different associated regions in an image. Furthermore, the region-based crack detection algorithm Faster R-CNN attained a higher accuracy when images were enhanced with directional lighting during all samples tested. The direction of highest accuracy was not consistent over samples, and is likely dependant on features such as crack location, width, orientation etc. This emphasises the importance of adaptive lighting: illuminating the surface with the most suitable conditions based on an initial observation of the feature or defect. This system represents the initial step in a fully automated and optimised concrete inspection system capable of defect capture, classification, localization and segmentation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.