Seeking a total automation of the existing industrial processes, manual product quality control systems have been gradually replaced by automated ones, to significantly improve efficiency and speed, and ultimately, increase industrial productivity. In this regard, an automated inspection system was developed in this work to detect and classify defects on the painted surfaces of Bosch Thermotechnology water heaters. This system comprised a deflectometry-based image acquisition module, two light deep learning models built and trained from scratch for defect detection and classification in the painted surfaces and a visual interface. The experimental results confirmed that: (1) deflectometry techniques were crucial for an accurate defect detection; (2) the two lightweight models – for detection and classification – rapidly achieved high accuracies, even in the testing stage, demonstrating their high performance regardless of their small size; (3) the developed system was able to correctly and quickly predict the status of a painted surface, and then successfully send this status information to a user-friendly visual interface, validating its suitability for an industrial setting. Overall, this system demonstrated great potential as a suitable alternative to the existing manual inspection of the painted surfaces of Bosch Thermotechnology water heaters.
PeresRSJiaXLeeJ, et al.Industrial artificial intelligence in industry 4.0 – systematic review, challenges and outlook. IEEE Access2020; 8: 220121–220139.
2.
GhobakhlooM. The future of manufacturing industry: a strategic roadmap toward industry 4.0. J Manuf Technol Manag2018; 29: 910–936.
3.
TorresYNadeauSLandauK. Classification and quantification of human error in manufacturing: a case study in complex manual assembly. Appl Sci (Switzerland)2021; 11: 1–23.
4.
BhattPMMalhanRKRajendranP, et al.Image-based surface defect detection using deep learning: a review. J Comput Inf Sci Eng2021; 21: 1–23.
5.
AkhtarSTandiyaAMoussaM, et al.An efficient automotive paint defect detection system. Adv Sci Technol Eng Syst2019; 4: 171–182.
6.
SinghSAKumarASDesaiKA. Comparative assessment of common pre-trained CNNs for vision-based surface defect detection of machined components. Expert Syst Appl2023; 218: 119623.
7.
SinghSADesaiKA. Automated surface defect detection framework using machine vision and convolutional neural networks. J Intell Manuf2023; 34: 1995–2011.
8.
Arnqvist NPNgendangenzwaBLindahlE, et al.Efficient surface finish defect detection using reduced rank spline smoothers and probabilistic classifiers. Econom Stat2021; 18: 89–105.
9.
LiYZhangY. Application research of computer vision technology in automation. In: 2020 international conference on computer information and big data applications (CIBDA). IEEE, 2020, pp. 374–377.
10.
HassaballahMAwadAI. Deep Learning in Computer Vision. 1st ed. Boca Raton, FL: CRC Press/Taylor and Francis, 2020, Epub ahead of print 23 March 2020. DOI: 10.1201/9781351003827.
11.
KimTHKimHRChoYJ. Product inspection methodology via deep learning: an overview. Sensors2021; 21: 1–19.
12.
MalamasENPetrakisEGMZervakisM, et al.A survey on industrial vision systems, applications and tools. Image Vis Comput2003; 21: 171–188.
13.
RenZFangFYanN, et al.State of the art in defect detection based on machine vision. Int J Precis Eng Manuf – Green Technol2022; 9: 661–691.
14.
Isasi-AndrieuAGarrote-ContrerasEIriondo-BengoaP, et al. Deflectometry setup definition for automatic chrome surface inspection. IEEE international conference on emerging technologies and factory automation, ETFA, 2017, pp. 1–4.
15.
ArnqvistNPNgendangenzwaBLindahlE, et al.Effective degrees of freedom for surface finish defect detection and classification. DOI: 0.48550/arXiv.1906.11904.
16.
GuanJFeiJLiW, et al.Defect classification for specular surfaces based on deflectometry and multi-modal fusion network. Opt Lasers Eng2023; 163: 107488.
17.
MolinaJSolanesJEArnalL, et al.On the detection of defects on specular car body surfaces. Robot Comput Integr Manuf2017; 48: 263–278.
18.
ChaiJZengHLiA, et al.Deep learning in computer vision: a critical review of emerging techniques and application scenarios. Mach Learn Appl2021; 6: 100134.
19.
AlzubaidiLZhangJHumaidiAJ, et al.Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. J Big Data2021; 8: 53.
20.
YamashitaRNishioMDoRKG, et al.Convolutional neural networks: an overview and application in radiology. Insights Imaging2018; 9: 611–629.
RawatWWangZ. Deep convolutional neural networks for image classification: a comprehensive review. Neural Comput2017; 29: 2352–2449.
23.
LuoCLiXWangL, et al. How does the data set affect CNN-based image classification performance? In: 2018 5th international conference on systems and informatics, ICSAI 2018; 2019, pp. 361–366.
24.
LeXMeiJZhangH, et al.A learning-based approach for surface defect detection using small image datasets. Neurocomputing2020; 408: 112–120.
25.
ZhangHWangDChenZ, et al.Adaptive visual detection of industrial product defects. PeerJ Comput Sci2023; 9: 1–23.
26.
Ruediger-FlorePGlattMHussongM, et al.CAD-based data augmentation and transfer learning empowers part classification in manufacturing. Int J Adv Manuf Technol2023; 125: 5605–5618.
27.
LiuJGuoFGaoH, et al.Defect detection of injection molding products on small datasets using transfer learning. J Manuf Process2021; 70: 400–413.
28.
YuDXuQGuoH, et al.An efficient and lightweight convolutional neural network for remote sensing image scene classification. Sensors2020; 20: 1999.
29.
JinTYanCChenC, et al.Light neural network with fewer parameters based on CNN for fault diagnosis of rotating machinery. Measurement (Lond)2021; 181: 109639.
30.
TuggenerLSchmidhuberJStadelmannT. Is it enough to optimize CNN architectures on ImageNet?Front Comput Sci2022; 4: 1041703.
31.
KornblithSShlensJLeQV. Do better ImageNet models transfer better? In: 2019 IEEE/CVF conference on computer vision and pattern recognition (CVPR). IEEE, 2019, pp. 2656–2666.
32.
YangJWSongDHAnHJ, et al.Classification of subtypes including LCNEC in lung cancer biopsy slides using convolutional neural network from scratch. Sci Rep2022; 12: 1–11.
33.
O’MahonyNCampbellSCarvalhoA, et al. Deep learning vs. traditional computer vision. In: Advances in intelligent systems and computing. Switzerland: Springer Verlag, 2019, pp. 128–144.
34.
ArmestoLTorneroJHerraezA, et al. Inspection system based on artificial vision for paint defects detection on cars bodies. In: 2011 IEEE international conference on robotics and automation. Shanghai, China, 2011, pp. 1–4. IEEE. DOI: 10.1109/ICRA.2011.5980570.
35.
ZhouQChenRHuangB, et al.DeepInspection: Deep learning based hierarchical network for specular surface inspection. Measurement (Lond)160, 2020; 107834. DOI: 10.1016/j.measurement.2020.107834.
36.
ChangFDongMLiuM, et al.A lightweight appearance quality assessment system based on parallel deep learning for painted car body. IEEE Trans Instrum Meas2020; 69: 5298–5307.
37.
BerwoMAFangYMahmoodJ, et al.FAECCD-CNet: fast automotive engine components crack detection and classification using ConvNet on images. Appl Sci2022; 12: 9713.
38.
UrbonasARaudonisVMaskeliunasR, et al. Automated identification of wood veneer surface defects using faster region-based convolutional neural network with data augmentation and transfer learning. Appl Sci (Switzerland)2019; 9: 4898.
39.
VenkateshSNSugumaranV. A combined approach of convolutional neural networks and machine learning for visual fault classification in photovoltaic modules. Proc IMechE, Part O: J Risk and Reliability2022; 236: 148–159.
40.
ParkJKKwonBKParkJH, et al.Machine learning-based imaging system for surface defect inspection. Int J Precis Eng Manuf – Green Technol2016; 3: 303–310.
41.
TianZQuPLiJ, et al. A survey of deep learning-based low-light image enhancement. Sensors2023; 23: 7763.
42.
MaoWLChiuYYLinBH, et al.Integration of deep learning network and robot arm system for rim defect inspection application. Sensors2022; 22: 3927.
