Manual crack inspection is labor-intensive and impractical at scale, prompting a shift toward AI-based segmentation methods. We present a novel crack segmentation model that leverages the Segment Anything Model 2 (SAM 2) through transfer learning to detect cracks on masonry surfaces. Unlike prior approaches that rely on encoders pretrained for image classification, we fine-tune SAM 2, originally trained for segmentation tasks, by freezing its Hiera encoder and FPN neck, while adapting its prompt encoder, LoRA matrices, and mask decoder for the crack segmentation task. No prompt input is used during training to avoid detection overhead. Our aim is to increase robustness to noise and enhance generalizability across different surface types. This work demonstrates the potential of foundational segmentation models in enabling more reliable and field-ready AI-based crack detection tools.
DOCUMENT
Masonry structures represent the highest proportion of building stock worldwide. Currently, the structural condition of such structures is predominantly manually inspected which is a laborious, costly and subjective process. With developments in computer vision, there is an opportunity to use digital images to automate the visual inspection process. The aim of this study is to examine deep learning techniques for crack detection on images from masonry walls. A dataset with photos from masonry structures is produced containing complex backgrounds and various crack types and sizes. Different deep learning networks are considered and by leveraging the effect of transfer learning crack detection on masonry surfaces is performed on patch level with 95.3% accuracy and on pixel level with 79.6% F1 score. This is the first implementation of deep learning for pixel-level crack segmentation on masonry surfaces. Codes, data and networks relevant to the herein study are available in: github.com/dimitrisdais/crack_detection_CNN_masonry.
DOCUMENT
An illustrative non-technical review was published on Towards Data Science regarding our recent Journal paper “Automatic crack classification and segmentation on masonry surfaces using convolutional neural networks and transfer learning”.While new technologies have changed almost every aspect of our lives, the construction field seems to be struggling to catch up. Currently, the structural condition of a building is still predominantly manually inspected. In simple terms, even nowadays when a structure needs to be inspected for any damage, an engineer will manually check all the surfaces and take a bunch of photos while keeping notes of the position of any cracks. Then a few more hours need to be spent at the office to sort all the photos and notes trying to make a meaningful report out of it. Apparently this a laborious, costly, and subjective process. On top of that, safety concerns arise since there are parts of structures with access restrictions and difficult to reach. To give you an example, the Golden Gate Bridge needs to be periodically inspected. In other words, up to very recently there would be specially trained people who would climb across this picturesque structure and check every inch of it.
LINK
This paper presents a proof of concept for monitoring masonry structures using two different types of markers which are not easily noticeable by human eye but exhibit high reflection when subjected to NIR (near-infrared) wavelength of light. The first type is a retroreflective marker covered by a special tape that is opaque in visible light but translucent in NIR, while the second marker is a paint produced from infrared reflective pigments. The reflection of these markers is captured by a special camera-flash combination and processed using image processing algorithms. A series of experiments were conducted to verify their potential to monitor crack development. It is shown that the difference between the actual crack width and the measured was satisfactorily small. Besides that, the painted markers perform better than the tape markers both in terms of accuracy and precision, while their accuracy could be in the range of 0.05 mm which verifies its potential to be used for measuring cracks in masonry walls or plastered and painted masonry surfaces. The proposed method can be particularly useful for heritage structures, and especially for acute problems like foundation settlement. Another advantage of the method is that it has been designed to be used by non-technical people, so that citizen involvement is also possible in collecting data from the field.
DOCUMENT
This paper outlines an investigation into the updating of fatigue reliability through inspection data by means of structural correlation. The proposed methodology is based on the random nature of fatigue fracture growth and the probability of damage detection and introduces a direct link between predicted crack size and inspection results. A distinct focus is applied on opportunities for utilizing inspection information for the updating of both inspected and uninspected (or uninspectable) locations.
DOCUMENT
Europe faces significant challenges in maintaining its aging infrastructure due to extreme weather events, fluctuating groundwater levels, and rising sustainability demands. Ensuring the safety and longevity of infrastructure is a critical priority, especially for public organizations responsible for asset management. Digital technologies have the potential to facilitate the scaling and automation of infrastructure maintenance while enabling the development of a data-driven standardized inspection methodology. This extended abstract is the first phase of a study that examines current structural inspection methods and lifecycle monitoring activities of the Dutch public and private entities. The preliminary findings presented here indicate a preference for data-driven approaches, though challenges in data collection, processing, personnel resources and analysis remain. The future work will experiment integrating advanced tools, such as artificial intelligence supported visual inspection, on the existing inspection datasets of these authorities for quantifying their readiness levels to the fully automated digital inspections.
DOCUMENT
Post-earthquake structural damage shows that wall collapse is one of the most common failure mechanisms in unreinforced masonry buildings. It is expected to be a critical issue also in Groningen, located in the northern part of the Netherlands, where human-induced seismicity has become an uprising problem in recent years. The majority of the existing buildings in that area are composed of unreinforced masonry; they were not designed to withstand earthquakes since the area has never been affected by tectonic earthquakes. They are characterised by vulnerable structural elements such as slender walls, large openings and cavity walls. Hence, the assessment of unreinforced masonry buildings in the Groningen province has become of high relevance. The abovementioned issue motivates engineering companies in the region to research seismic assessments of the existing structures. One of the biggest challenges is to be able to monitor structures during events in order to provide a quick post-earthquake assessment hence to obtain progressive damage on structures. The research published in the literature shows that crack detection can be a very powerful tool as an assessment technique. In order to ensure an adequate measurement, state-of-art technologies can be used for crack detection, such as special sensors or deep learning techniques for pixel-level crack segmentation on masonry surfaces. In this project, a new experiment will be run on an in-plane test setup to systematically propagate cracks to be able to detect cracks by new crack detection tools, namely digital crack sensor and vision-based crack detection. The validated product of the experiment will be tested on the monument of Fraeylemaborg.
Lean Production (LP) can be regarded as a design approach in search of a theoretical foundation. In this paper we show that Lowlands’ Sociotechnical Design Theory (STSL) could function as such a foundation. To reach this goal, we first describe STSL as a system theoretical reformulation of Original Sociotechnical Theory (OSTS). Then, we introduce the Toyota Production System as the origin of LP and the challenge it poses for the academic field of organization design. This academic field should (1) assess LP’s success, (2) generalize it by embedding it in more abstract concepts and theories in order to be able to (3) re-specify it for different manufacturing and non-manufacturing contexts. Next, we give an exposition of STSL as a structural design approach based on developments in system theory. At last, we reformulate lean production in STSL terms and so show that LP is a subcase within the more general theory of STSL. We discuss the merits of both approaches and clarify some misunderstandings of lean both outside and inside the lean community. Embedding LP in the more general language of STSL should enable us to discover similarities and differences, to start a process of mutual learning, to integrate diverse design approaches in a theory of organizational design and to add content to redesign proposals of for example the health care system as proposed by Porter and Teisberg (2006) and Christensen et al. (2009). We quote extensively from the lean literature (to convince our sociotechnical friends) and embed both STSL and LP in the broader literature on organization design. We hope this adds a new perspective to the one given in the Operations Management literature on LP. Again, mutual learning is the goal.
DOCUMENT
During the 2015 Gorkha earthquake of 7.8 Mw that hit Kathmandu Valley, Nepal, numerous Nepalese Pagodas suffered extensive damage while others collapsed. Risk reduction strategies implemented in the region focused on disassembling historical structures and rebuilding them with modern material without in depth analysis of why they suffer damage and collapse. The aim of this paper is to evaluate the effectiveness of low-cost, low-intervention, reversible repair and strengthening options for the Nepalese Pagodas. As a case study, the Jaisedewal Temple, typical example of the Nepalese architectural style, was investigated. A nonlinear three-dimensional finite element model of the Jaisedewal Temple was developed and the seismic performance of the temple was assessed by undertaking linear, nonlinear static and nonlinear dynamic analyses. Also, different structural intervention options, suggested by local engineers and architects working in the restoration of temples in Nepal, were examined for their efficacy to withstand strong earthquake vibrations. Additionally, the seismic response of the exposed foundation that the Nepalese Pagodas are sitting on was investigated. From the results analysis, it was found that pushover analysis failed to capture the type of failure which highlights the necessity to perform time-history analysis to accurately evaluate the seismic response of the investigated temple. Also, stiffening the connections along the temple was found to enhance the seismic behaviour of the temple, while strengthening the plinth base was concluded to be insignificant. Outputs from this research could contribute towards the strategic planning and conservation of multi-tiered temples across Nepal and reduce their risk to future earthquake damage without seriously affecting their beautiful architectural heritage.
DOCUMENT
Few people I know act likea magnet in the way Laura does. If you hear her speak, see her move, enjoy her smile - you can't help but want to be a part of what makes her heart beat faster. She radiates passion for her dream topic biomimicry and before you know it you're caught in that dream. From the day I met her, I was captivated by her enthusiasm and knowledge about this topic. In fact, meeting Laura made me decide to join the university as I thought: if people like Laura work at THUAS, I want to be a part of this organization'. Over the years I have seen her finish the Msc in biomimicry at Arizona State University followed by a PhD at TUDelft. And all that next to her full time job as a teacher. It's only a miracle that she still found the time to go outdoors and be in Nature. But luckily, she did as this is what nourishes her - and that nourishment is brought into the classroom affecting generations of students. I am very proud of how she builds her tribe just like Nature does; grassroots from the bottom up, not supported but also not inhibited by formal structures. In that way she truly acts as a bridge helping others to tap into Nature's wisdom. This morning I harvested the remaining vegetables from my garden and turned them into lunch. It's the second year I grow vegetables and it feels like I am only at the beginning of learning to collaborate with Nature. In Spring and Summer, Ihave witnessed in awe how seeds become seedlings which then grow into mature plants carrying fruit. The sheer wonder of Nature never ceases to amaze me, and my garden is only an attempt to be more aware of seasonal rhythms. It's Autumn right now, a time of year that invites us to go inside, reflect and let go of old baggage that no longer serves us. We'll be approaching the stage of wintering soon in which our inner journey will benefit from the darkness of wintertime introspection, along with the space to process the old, integrate learnings, and then germinate the new. Over the course of her career, Laura has gone through these seasonal cycles - reinventing herself in the past decade as a teacher, researcher and regenerative leader. One of Laura's many qualities is that she embodies three leadership characteristics derived from Nature. First, she acknowledges the importance of interconnection. Many times, we think of Nature as being separate from us, but in reality we humans are Nature. Connection with Nature enables us to think within systems and understand that we can't direct the system, but instead we're all part of multiple systems. Second, sensing the system and our part in it builds resilience. Even if things don't go as we expected or imagined, rather than reacting, we can step back and engage with more insight. Laura's adaptability to a system's needs while spotting opportunities to crack it open, is admirable. As the system is always in evolution, so is she - remarkably receptive to change even in the final stages of her career. Third, Laura creates space for people to develop and thrive, acting as multipliers of her vision and love for the natural world. In her leadership she embodies the ideal elder while being able to perceive the world through the eyes of a child - with continuous wonder for how life unfolds. This book is a bricolage of Laura's post-doc research conducted the past two years. In it you will find an array of fascinating reads and tools that help you deepen your practice as a biomimicry professional. The book is a community effort integrating tools Laura has co-created with her ecosystem as well as more in-depth readings written by some of the talents she has nourished over time. I wish for you to enjoy this careful curation of both practical as well as more conceptual contributions. May it inspire your own thriving in bringing Nature based wisdom to the core of our daily lives.
DOCUMENT
