Do the rebar scaffolding pipes at the construction site need to be counted one by one? It can be done with just a smartphone!
Thread Content
In recent years, Shanghai Construction Engineering Group has actively explored the technical applications of artificial intelligence in the construction industry. It developed its own WeChat mini-program called “Reinforcement Bar Counting Cloud”, which utilizes computer vision, deep learning, and 5G technology; by taking a photo with a smartphone, it is possible to automatically count the number of reinforcement bars/steel pipes in just 2.3 seconds. To date, this mini-program has been optimized to version V5.0 and is being widely used in key projects under construction by Shanghai Construction Group, such as the Shanghai Grand Opera House, Xujiahui Center, and Expo Hotel. It is also available for use by external enterprises, research institutions, universities, and other organizations. The mini-program was called approximately 40,000 times, involving over 300 actual engineering projects, and around 3.5 million steel bars/steel pipes were processed using cloud-based counting.
“The WeChat mini-program “Reinforcement/Steel Pipe Dot Counting” was independently developed by Shanghai Construction Engineering Group No.4. It was officially launched on October 26, 2020. The core algorithms for identifying reinforcement/steel pipes and counting their dots were all developed by Group No.4 itself, and it is now available for use by external organizations.
A FPN-based multi-scale feature fusion convolutional neural network model. Steel bars in images present complex visual characteristics such as small object size, object overlap, varying object dimensions, and diverse shooting backgrounds. To improve the ability of convolutional neural network models to extract features of steel bars, Si Jian Group developed a FPN-based multi-scale feature fusion convolutional neural network model that enhances the accuracy of detecting small overlapping objects by fusing multi-scale feature maps of the target objects. In the newly released V5.0 version of the mini-program, the accuracy rate for detecting rebar clouds is close to 95%, while the accuracy rate for detecting steel pipe clouds is close to 98.5%.
It’s truly high-tech. At first, I thought there were QR codes attached to each rebar.
Behind, the application of 5G technology is powerful