A Deep-Learning-based 3D Defect Quantitative Inspection System in CC Products Surface

Abstract

To create an intelligent surface region of interests (ROI) 3D quantitative inspection strategy a reality in the continuous casting (CC) production line, an improved 3D laser image scanning system (3D-LDS) was established based on binocular imaging and deep-learning techniques. In 3D-LDS, firstly, to meet the requirements of the industrial application, the CCD laser image scanning method was optimized in high-temperature experiments and secondly, we proposed a novel region proposal method based on 3D ROI initial depth location for effectively suppressing redundant candidate bounding boxes generated by pseudo-defects in a real-time inspection process. Thirdly, a novel two-step defects inspection strategy was presented by devising a fusion deep CNN model which combined fully connected networks (for defects classification/recognition) and fully convolutional networks (for defects delineation). The 3D-LDS’ dichotomous inspection method of defects classification and delineation processes are helpful in understanding and addressing challenges for defects inspection in CC product surfaces. The applicability of the presented methods is mainly tied to the surface quality inspection for slab, strip and billet products.

Advanced Computational Intelligence for Object Detection, Feature Extraction and Recognition in Smart Sensor Environments

“…We kindly encourage the scientific community to adapt this data set in the research on camera pose regression methods. In [ 8 ], a model to detect surface regions of interest (ROI) in 3D was presented. As a processing mechanism, a deep convolutional neural network (CNN) modeling mechanism was adapted with the Adam training algorithm.…”

Quality Control of PET Bottles Caps with Dedicated Image Calibration and Deep Neural Networks

“…It is related to the reduction of image processing parameters that should be set for proper product control, as well as the possibility of easy adaptation of existing algorithms or settings to other formats of a given product or other production lines. Therefore, the classical methods of image processing are increasingly being replaced by methods of deep-based neural networks [ 9 , 10 , 11 , 12 ], and the necessity of manual parameters setting in classic image processing algorithms turns into the model training process, i.e., collecting the appropriate number of examples of tested products (as well as examples with defects in the case of supervised learning methods). Collecting a sufficient number of examples with defects or faults can be cumbersome in some real cases.…”

FPGA-Based Acceleration on Additive Manufacturing Defects Inspection

“…Chua et al [ 5 ] reviewed current process monitoring/control systems for metal AM and proposed a comprehensive real-time inspection method and a closed-loop monitoring system to improve the quality of AM printed parts. To ensure the internal quality while improving the efficiency of AM product inspection, deep learning is commonly applied in both industry and academia [ 6 , 7 , 8 ]. Typically, the deep learning-based defects inspection involves three steps: image acquisition, image classification, and defects localization.…”