Trendspek, in collaboration with Datature, has developed a proof-of-concept tool to expedite concrete crack detection in infrastructure using advanced computer vision and AI machine learning.
In an exciting first step to introduce AI to built asset inspection, Trendspek, a 3D asset management and reporting platform, and Datature, an MLOps computer vision platform, have developed a proof-of-concept for an automated structural crack detection tool.
These machine learning models are trained to identify concrete cracks on large structures up to 5 faster than manual analysis, allowing asset owners to vastly streamline defect detection and maintenance.
How it works
Datature is an end-to-end MLOps platform designed to allow users to upload their own data to create their own custom computer vision without coding experience.
Using Datature’s Nexus platform, Trendspek imported drone image data and used Nexus’ in-app tools, including in-house Annotator, Paintbrush, Intellibrush and Automation tool, to precisely label polygons around each structural crack, training the semantic segmentation model within 24 hours.
To quickly identify the small cracks present in the large, high-quality 8K images, Trendspek used Datature Nexus’ Sliding Window feature, designed specifically to aid small object detection, as well as Image augmentation techniques to help models become more robust against varying weather conditions.
The prediction results from the Datature models were integrated with Trendspek Precision Reality Twin (PRT) 3D models and then rendered on Trendspek’s Precision Asset Intelligence platform, successfully identifying over 8000 cracks across 500 images to provide fast, accurate insights.
Within this first experiment, we were able to collaboratively produce a working proof-of-concept model for an AI tool that can detect concrete cracks with unrivalled speed.
Josh Sinclair, Head of Product at Trendspek says:
"Our Processing team spent a week highlighting 8019 cracks across 500 images with the Datature Nexus platform. This data was used to train a DeepLapV3 neural network with the ability to detect cracks in concrete.
By leveraging Datature's Active Learning capabilities, we're looking at reducing a 100-hour process down to 3 hours, plus time for an engineer to quality control the model.
This provides an enormous opportunity to bring AI capabilities for all kinds of assets, and to expand our scope to train new models and identify other structural defects.
Trendspek PRT integrated with Datature training model
What comes next?
While still in its development, this integration aims to allow users to leverage the efficiency of Datature’s automated computer vision solutions with AI-powered analysis, offering users a tool to speed up crack defect identification by up to 5 times.
While manual analysis remains crucial for infrastructural maintenance, the potential for this technology to enhance existing inspection processes is enormous.
Derek Feebrey, CEO at Trendspek says:
“As we develop this integration, it opens up an amazing opportunity for our clients to easily train their own AI models. This will be a real game-changer for infrastructure owners to streamline their defect identification and maintenance."
Keechin Goh, CEO at Datature says:"I am thrilled to share our enthusiastic support for Trendspek’s mission to transform asset management through their groundbreaking Precision Asset Intelligence 3D visualization software. Our collaboration has birthed an advanced machine-learning pipeline dedicated to structural crack detection, significantly enhancing the capability of Trendspek’s software to identify and segment cracks with unparalleled efficiency."
FAQs
When will this AI integration be available for Trendspek users?
This proof-of-concept will allow us to accurately scope future projects. It is unclear when this feature will be generally available, however, if you feel you have a use case for this technology please get in touch to discuss the potential for further proof of concepts.
Will it cost more?
There will likely be costs associated to access Trendspek’s AI-powered features, which we will provide more detail on in due course.
Where can I see the latest Trendspek product updates?
While logged into our platform, click the Chatbot icon in the bottom right corner of the screen. Follow the prompts to check the progress of priority items in our product roadmap.
About Trendspek
Trendspek is the world’s leading Precision Asset Intelligence software that empowers you to virtually inspect and monitor infrastructural assets safely, securely and remotely. Transforming tens of thousands of high-quality images into exact 3D replicas, Precision Reality Twins, users can streamline defect identification and conduct thorough condition assessments from their desks. This helps enterprises to accelerate and cut down costs for asset markup, monitoring, and maintenance planning.
About Datature
Datature is a leading end-to-end MLOps platform that allows teams and enterprises to build computer vision models without a single line of code. Teams can manage datasets, annotate, generate synthetic data, train and deploy - all in a single, secure cloud-based platform. With the rise of citizen data scientists, deep tech companies, and enterprises looking to adopt deep-learning - Datature equips these startups / experts with the tools required to build their own capabilities easily within weeks