Automated Surface Inspection of Micro Parts

Scholz‐Reiter, Bernd; Thamer, Hendrik; Lütjen, Michael

doi:10.1063/1.3510557

Cited by 3 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The workbench represented in Fig. 3 is composed of Minitec aluminium profiles (1) making it easily adaptable to new products, bolted steel sheets to hold the scanner (2) and hold a laser sensor (3). Space is also allocated to store the unused door panel housings (6) and to hold the computer (7).…”

Section: Resultsmentioning

confidence: 99%

“…The companies' efforts include implementing lean and systematic methodologies and increasingly reliable and advanced control systems. Optical systems emerge that allow checking the execution of tasks and the status of products using various technologies such as infrared light, diffusion, and fibres, among others [3]. In a fully automated process, the use of measuring devices is essential.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Development of Quality Assurance System in the Automotive Industry

Gonçalves,

Campilho,

Prakash

2023

Advances in Science and Technology

View full text Add to dashboard Cite

The Automotive Industry is in constant evolution, thanks in part to its high level of production and customer demands for quality assurance. New processes, equipment and products are constantly emerging, leading to new vehicles launched to the public. These processes and equipment need adequate monitoring tools and forms of quality assurance, and it is the responsibility of the engineering department to look for economically viable and efficient ways of guaranteeing the quality of products in an industry where quality itself is considered one of its pillars. This work aims to develop and validate a solution for quality control in the automotive industry while taking full advantage of the technological factors provided by the equipment used. It deals with researching and applying a three-dimensions (3D) scanner in product quality monitoring tasks of a door panel manufacturing process. Pilot studies were carried out to assess the real capabilities of the equipment and determine problems that could arise when scaling the process for a real production of car door panels. The implementation and final results are also described in this work, showing that it is possible to develop a flexible equipment with multiple applications and apply it in the automotive industry.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Development of Quality Assurance System in the Automotive Industry

Gonçalves,

Campilho,

Prakash

2023

Advances in Science and Technology

View full text Add to dashboard Cite

show abstract

“…For inference times, all CNNs were capable of running in real-time, with the fastest one being Resnet18 with an inference time of 0.057 ± 0.016 seconds, and the slowest one being VGG19, with an inference time of 0.269±0.002 seconds. Deep CNN [29] Statistical features [23] SIFT and ANN [24] Weibull [27] TPR (%) As for VGG16, the one selected to be the feature extraction component of Auto-Classifier, it had an inference time of 0.225±0.002, which is enough for detecting surface defects in real-time. CNN-fusion, in a serial manner, has an f time of 1.152 seconds, which is not suitable for real-time.…”

Section: Resultsmentioning

confidence: 99%

“…However, their results fall short when compared to deep learning techniques. In [23], a method based on LBP achieved 95.9% accuracy, and [24], that achieved a 98.2% accuracy by using EANT2, a neuroevolution method to develop artificial neural networks for classification purposes.…”

Section: Related Workmentioning

confidence: 99%

“…However, such systems are scarce due to difficulties in acquiring real data to train Artificial Intelligence (AI) algorithms, and because industry floors require continuous changes, meaning that having controlled and unchanged environments is challenging. Traditional methods often rely on extracting hand-craft features from images, in order to represent defects and anomalies [23,24] and can be categorized into: statistical, structural or filter based. These methods are capable of detecting and or segmenting defects in images if they are acquired in a controlled environment [31].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Auto-Classifier: A Robust Defect Detector Based on an AutoML Head

Lopes,

Alexandre

2020

Preprint

View full text Add to dashboard Cite

The dominant approach for surface defect detection is the use of hand-crafted feature-based methods. However, this falls short when conditions vary that affect extracted images. So, in this paper, we sought to determine how well several state-of-the-art Convolutional Neural Networks perform in the task of surface defect detection. Moreover, we propose two methods: CNN-Fusion, that fuses the prediction of all the networks into a final one, and Auto-Classifier, which is a novel proposal that improves a Convolutional Neural Network by modifying its classification component using AutoML. We carried out experiments to evaluate the proposed methods in the task of surface defect detection using different datasets from DAGM2007. We show that the use of Convolutional Neural Networks achieves better results than traditional methods, and also, that Auto-Classifier out-performs all other methods, by achieving 100% accuracy and 100% AUC results throughout all the datasets.

show abstract