Computer Vision Rebar Arrangement Verification Using Gap Mask Classification

The patent replaces manual visual inspection with an automated computer vision system that captures images of rebar arrangements and uses deep learning models to automatically detect and classify space intervals. This substitution of mechanical/manual inspection with automated optical-digital systems resolves the contradiction by significantly improving detection accuracy while maintaining reasonable system complexity through the use of standard imaging equipment and pre-trained neural networks.

The patent creates a digital copy of the physical rebar arrangement through image capture and processing. By generating mask regions that represent spatial relationships between rebars in the image domain, the system enables accurate measurement and classification of intervals without physically measuring each rebar. This copying approach improves detection precision while avoiding the complexity of direct physical measurement systems.

The patent employs pre-trained deep learning models that have been previously trained on large datasets of rebar images. This preliminary training action allows the inspection system to quickly process new images without requiring extensive computational resources during actual inspection. The pre-trained models enable high productivity by performing rapid inference, while the computational energy consumption is optimized by reusing the trained weights rather than retraining for each inspection task.

The patent extracts only the critical information needed for inspection by generating mask regions that highlight space intervals between rebars. Instead of processing entire images or analyzing every pixel, the system extracts and focuses on the specific spatial relationship data, improving inspection efficiency while reducing unnecessary computational energy consumption on irrelevant image data.