On-line machine vision system for detect split defects in sheet-metal forming processes

Gayubo, Fernando; González, José Luís; Fuente, Eusebio de la; Miguel, Fernando Manero; Perán, José R.

doi:10.1109/icpr.2006.902

Cited by 21 publications

(11 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this is done in an off-line procedure, so adaptivity relates to the interaction level with the user rather than adapting the classifiers with new images. In [12] an on-line machine vision system is presented for defect detection in sheet-metal forming processes, however the word 'on-line' is related to the classification decisions rather than applying an on-line adaptation procedure for the classifiers. Sometimes, the image classifiers are updated, but this is usually handled by either an update of the reference data base (for nearest neighbor based approaches) or by a re-training step on the whole set of images seen so far [13].…”

Section: Motivationmentioning

confidence: 99%

On-line evolving image classifiers and their application to surface inspection

Lughofer

2010

Image and Vision Computing

View full text Add to dashboard Cite

Section: Motivationmentioning

confidence: 99%

On-line evolving image classifiers and their application to surface inspection

Lughofer

2010

Image and Vision Computing

View full text Add to dashboard Cite

“…There are several sensors used to be implemented in the final designed jig. The sensors suitable are machine vision system [2], fibre optic laser scanning system [3] and piezoelectric sensor [4].…”

Section: Figure 3: Locking Of Seal With Wrong Orientation Of Jawmentioning

confidence: 99%

Design of Reverse and Missing Jaw Detector Jig using Analytic Hierarchy Process (AHP)

Noor¹,

Jafar²,

Bakar³

et al. 2019

IJRTE

View full text Add to dashboard Cite

A seal manufacture factory located in Ipoh, Perak produces two types of seal available in the factory which are metal and plastic seals. Due to manual assembly, human operator tends to get fatigue for doing repetitive task in one long shift. When this occurs, minor error such as misplace or wrong orientation of jaw will occur. Moreover the jaw sometimes can be missing during transportation of work – in – progress to another process. The objective is to design a jig that have the ability to detect reverse and missing jaw so that the finish product which has the quality problem will be prevented from being shipped to customers. The jig must also have a flexibility to be used for other plastic seal product. For methodology, each and every stage was explained in detail. In result and discussion section, House of Quality (HOQ) was used to determine the highest criteria weightage. The important criteria which will be carried forward to Analytical Hierarchy Process (AHP) are 15.9% on flexibility and ease of maintenance, 15.4% on both ease of use and longevity of jig usage, lastly 15.2% for accuracy checking. The result obtain after AHP computation are Design 1 with 20.7%, Design 2 with 13.7%, Design 3 with 22%, Design 4 with 27.1% and Design 5 with 16.5%. Design 4 is selected due to the highest weightage after computation. Final selected design was justified according to characteristic needed in objective.

show abstract

“…However, as Machine Learning technology moves from research laboratories to practical applications such as Machine Vision, a range of issues arise concerning how humans relate to, and interact with such systems [9] [6]. Not only does this question the feasibility, or even relevance of considering "cleaned" data sets, there is an increasing demand for systems to operate in situations where off-line batch-mode processing is not appropriate [7]. This can occur if data is hard, time-consuming or costly to obtain, or if the underlying processes change fairly rapidly, requiring re-configuration.…”

Section: Introductionmentioning

confidence: 99%

On Human-Machine Interaction during Online Image Classifier Training

Lughofer

Smith

Caleb-Solly

et al. 2008

2008 International Conference on Computational Intelligence for Modelling Control &Amp; Automation

View full text Add to dashboard Cite

This paper considers on a number of issues that arise when a trainable machine vision system learns directly from humans, rather than from a "cleaned" data set, i.e. data which is perfectly labelled with complete accuracy. This is done within the context of a generic system for the visual surface inspection of manufactured parts, however, the issues treated are relevant not only to wider computer vision applications, but also to classification more generally. Some of these issues arise from the nature of humans themselves: they will be not only internally inconsistent, but will often not be completely confident about their decisions, especially if they are making decisions rapidly. People will also often differ systematically from each other in the decisions they make. Other issues may arise from the nature of the process, which may require the machine learning to have the capacity for real-time, online adaptation in response to users' input. It may be that the users cannot always provide input to a consistent level of detail. We describe how all of these issues may be tackled within a coherent methodology. Using a range of classifiers trained on real data sets from a CD imprint production process, we will present results which show that most of these issues may actually lead to improved performance.

show abstract

On-line machine vision system for detect split defects in sheet-metal forming processes

Cited by 21 publications

References 3 publications

On-line evolving image classifiers and their application to surface inspection

On-line evolving image classifiers and their application to surface inspection

Design of Reverse and Missing Jaw Detector Jig using Analytic Hierarchy Process (AHP)

On Human-Machine Interaction during Online Image Classifier Training

Contact Info

Product

Resources

About