Rapid estimation method of the residual mortar of recycled aggregates by object detection models

Removing the attached mortar from recycled aggregates through crushing and grinding is a mainstream method to improve the quality of recycled aggregates. Due to the heterogeneity of recycled aggregates, it is necessary to adjust process parameters in real-time during the grinding process to achieve product quality control . This requires the product inspection stage to accurately distinguish between recycled aggregates and attached mortar, and to quickly evaluate the effects of crushing and grinding. This study is based on the YOLO v7 object detection model. By improving the network structure, the detection efficiency of the model was enhanced, and the real-time calculation of the relative content of mortar and recycled aggregate was achieved. Comparing the detection results of computer vision method with traditional methods based on moisture loss rate, the results show that this rapid detection method is comparable to traditional methods in terms of effectiveness and accuracy, making it an efficient evaluation method for product quality in the grinding process. • The image method allowing for rapid detection of mortar and recycled aggregates. • The mortar area measured by the image method is linearly related to the mortar content. • The accuracy of the image method is comparable to those of traditional methods.