Machine learning software for inspecting building facades

برمجيات التعلم الآلي لفحص واجهات المباني

Machine learning and artificial intelligence have been in the construction industry to make some real impact ripples.

After nearly two years of development in Thornton Tomasetti’s CORE lab,

Engineers have created a computer vision and machine learning algorithm known as Thornton Tomasetti Damage Detector (T2D2).

Which can help to identify the damage to the outbuilding through videos or photos.

T2D2 Software as a Service (SaaS) aims to quickly detect any subtle anomalies that might go unnoticed during manual interface checking, which is a tedious and difficult process.

The system will also be directed to some extent, and engineers will review the discoveries made by the T2D2,

Which will be flexible enough to search for conditions across different types of materials and structure.

Machine learning software for inspecting building facades

 

It can be used for all types of structures, but the main objective is to screen older structures that require periodic inspections.

To get an idea of the tool’s effectiveness, the team reviewed hundreds of drone images collected over decades of building inspections.

The results were very promising and they were able to assist human engineers before they approached any building for inspection.

T2D2 typically works in several steps.

 

Machine learning software for inspecting building facades

 

How T2D2 works

First, by using a cache of knowledge gained through thousands of pre-annotated images,

It quickly detects the geometry of the structure and identifies the types of materials.

Then it makes a pass to detect the possible damage to which the type of material is susceptible.

Finally, it produces a bolder version of the image, labeling potential damage and highlighting it for review.

The T2D2 was initially trained primarily on concrete structures, but the capabilities have now been expanded.

The tool can now identify and classify damage to masonry, brick, stucco, and other commonly used materials.

Due to the inclusion of machine learning elements, the algorithm gets better every time it is applied.

 

Machine learning software for inspecting building facades

 

Moreover, the engineers plan to use reinforcement learning to improve the models.

However, the engineer will be able to discern false positives and negatives, which helps to enhance results.

With T2D2, the way to inspect buildings is redefined.

Thornton Tomasetti engineers work in partnership with drone surveying companies to provide a comprehensive façade inspection service.

Building owners typically only perform façade inspections to meet regulatory requirements,

but the relatively low cost of using T2D2 combined with drones could change that.

Frequent inspections can mean building owners can make small repairs to prevent larger problems while saving millions in repairs.

 

For more architectural news

 

Types advantages and disadvantages of cantilever bridges

Leave a Reply

Your email address will not be published. Required fields are marked *