Three types of inspection complexity were tested on an inspection task using both industrial and student subjects. Items inspected varied with regard to the number of different fault types (two, four, or six), whether the inspecting standards for each fault type were the same or different, and whether faults occurred anywhere on the item or only on specific sub-areas. Number of fault types had a large effect on the search component of the task. The effect of same or different standards was largely confined to the decision-making component. There was no effect of faults being distributed across the whole or part of the item. The 18 industrial quality-control personnel were not significantly different in performance from the 48 student subjects tested.
Small batch manufacturing typical of modern production systems requires improved visual inspection of novel products for many defect types. Because each item inspected is different, inspectors cannot develop product-specific techniques to speed performance using traditional direct viewing. We performed two experiments to evaluate the use of binocular rivalry as an alternative to direct viewing in stereo viewing of product pairs. Both free stereo viewing (i.e., with no optical aids) and viewing with a stereoscope led to large increases in performance of searching for targets in extended symbol arrays. For increased target set size there was an expected increase in inspection time for direct viewing but no corresponding increase for stereo viewing. Binocular rivalry appears to be a promising, though difficult, tool for facilitating visual inspection, as it both improves performance and appears to allow parallel processing of targets.
A series of experiments was conducted to clarify the relationship between the time required and visual process for counting dots projected onto a screen. The eye movement of the subject on each experimental trial was recorded by the electro-oculograph method and eye mark recorder. According to the experimental results, the subject used two information processes depending on the number of dots and symmetry of patterns. Both latency and movement time depend on the configuration of dots in the sample. An excess number of saccades is one of the causes for miscounting. A model for estimating counting time was built on the basis of the findings obtained. The time estimated by the model, if the maximum number of saccades does not exceed the number of dots on each sample, may be taken as optimum counting time.
Objective: Sandia National Laboratories conducted an experiment for the National Nuclear Security Administration to determine the reliability of visual inspection of precision manufactured parts used in nuclear weapons. Background: Visual inspection has been extensively researched since the early 20th century; however, the reliability of visual inspection for nuclear weapons parts has not been addressed. In addition, the efficacy of using inspector confidence ratings to guide multiple inspections in an effort to improve overall performance accuracy is unknown. Further, the workload associated with inspection has not been documented, and newer measures of stress have not been applied. Method: Eighty-two inspectors in the U.S. Nuclear Security Enterprise inspected 140 parts for eight different defects. Results: Inspectors correctly rejected 85% of defective items and incorrectly rejected 35% of acceptable parts. Use of a phased inspection approach based on inspector confidence ratings was not an effective or efficient technique to improve the overall accuracy of the process. Results did verify that inspection is a workload-intensive task, dominated by mental demand and effort. Conclusion: Hits for Nuclear Security Enterprise inspection were not vastly superior to the industry average of 80%, and they were achieved at the expense of a high scrap rate not typically observed during visual inspection tasks. Application: This study provides the first empirical data to address the reliability of visual inspection for precision manufactured parts used in nuclear weapons. Results enhance current understanding of the process of visual inspection and can be applied to improve reliability for precision manufactured parts.
Numerous presentations and articles on manual inspection of pharmaceutical drug products have been released, since the pioneering articles on inspection by Knapp and associates Knapp and Kushner (J Parenter Drug Assoc 34:14, 1980); Knapp and Kushner (Bull Parenter Drug Assoc 34:369, 1980); Knapp and Kushner (J Parenter Sci Technol 35:176, 1981); Knapp and Kushner (J Parenter Sci Technol 37:170, 1983). This original work by Knapp and associates provided the industry with a statistical means of evaluating inspection performance. This methodology enabled measurement of individual inspector performance, performance of the entire inspector pool and provided basic suggestions for the conduct of manual inspection. Since that time, numerous subject matter experts (SMEs) have presented additional valuable information for the conduct of manual inspection Borchert et al. (J Parenter Sci Technol 40:212, 1986); Knapp and Abramson (J Parenter Sci Technol 44:74, 1990); Shabushnig et al. (1994); Knapp (1999); Knapp (2005); Cherris (2005); Budd (2005); Barber and Thomas (2005); Knapp (2005); Melchore (2007); Leversee and Ronald (2007); Melchore (2009); Budd (2007); Borchert et al. (1986); Berdovich (2005); Berdovich (2007); Knapp (2007); Leversee and Shabushing (2009); Budd (2009). Despite this abundance of knowledge, neither government regulations nor the multiple compendia provide more than minimal guidance or agreement for the conduct of manual inspection. One has to search the literature for useful information that has been published by SMEs in the field of Inspection. The purpose of this article is to restate the sound principles proclaimed by SMEs with the hope that they serve as a useful guideline to bring greater consistency to the conduct of manual inspection.
With the advent of Industry 4.0, the use of new technologies, robotization and advanced manufacturing has been extended to the agricultural sector, with the aim of increasing productivity, reducing environmental impacts, increasing profits and improving the quality of products, giving rise to the terms Precision Agriculture, Agribusiness 4.0, Agriculture 4.0 and Agroindustry 4.0. If on the one hand much is being said about the adoption of new technologies in the stages of land preparation, planting and harvesting, on the other hand very little is said about the processing of agricultural products using, for example, automated systems for visual inspection of quality. This work aims to investigate the different approaches for automatic visual inspection of grains quality proposed in the last decade and present a discussion about how these approaches are inserted in the context of these new productive processes of modern agriculture, as well as the positive aspects and the limitations found for their uses.
Dynamic visual inspection (DVI) occurs when objects on a moving conveyor are being individually examined for compliance with specifications. This study investigated the effect of the conveyor velocity and object interspacing on DVI performance and eye-motion behavior during this task. Nine combinations of these object presentation factors were examined where three pairs of these combinations each had a constant throughput rate (or exposure time). Horizontal eye-motion measurements were made through electrooculography recordings, and these measurements were analyzed to separately distinguish the visual acquisition time from the visual tracking time. Both of these time values tended to vary predictably with the available exposure time and consistently between people as a visual behavior strategy. Inspection errors were found to be highly correlated with the visual behavior.
The paper describes an Automated Visual Inspection - AVI- System embedded in the control of the production process of complex mechanical pieces. This implies integration of NDT systems (Automated Visual Inspection system, Optical Roughness assessment system and Laser measurement system, performing over CAD data from pieces), quality assessment system, production process's assessment system and tracking system. ; The authors want to acknowledge The DG XII of the Commission of the European Communities for SMARTMEC Project's financing within the BRITE EuRam Programme, as well as the Spanish government for CICYT's support .
Thirty-nine subjects participated in an experiment designed to evaluate the influence of the reflective-impulsive cognitive style on visual inspection. The Matching Familiar Figures Test (MFFT) was used to classify subjects as reflectives (longer times, fewer errors), impulsives (shorter times, more errors), fast-accurates (shorter times, fewer errors), and slow-inaccurates (longer times, more errors). Following administration of the MFFT, subjects participated in a laboratory visual inspection task. Results from the inspection task indicate that the MFFT groups coalesced along an accuracy rather than a speed dimension. The more accurate groups (reflectives and fast-accurates) were significantly faster than the inaccurates (impulsives and slow-inaccurates) in detecting certain flaws, and they made fewer size-judgment errors. However, the inaccurates detected more flaws, (i.e., made fewer search errors) than did the accurates. These results are interpreted in terms of the possible cognitive styles affecting inspection performance.
Background: Cervical cancer is a major public health problem especially in developing countries. It can be prevented through implementation of routine screening program. There are different screening methods but their efficacy are still questionable. So the purpose of this study is to evaluate the efficacy of visual inspection of cervix with acetic acid and colposcopy to detect precancerous lesion in women with clinically unhealthy or abnormal cervix.Methods: Forty patients with abnormal cervix (35) and abnormal pap smear results (5) were enrolled for the study in outpatient department of Kathmandu Model Hospital. Patients were evaluated with visual inspection of cervix with acetic acid and colposcopy in the same sitting. Cervical punch biopsy were taken from suspected lesion or from four quadrant if colposcopy findings were normal and sent for histopathological examination. The finding of visual inspection of cervix with acetic acid and colposcopy were correlated with histopathological finding and compared with each other.Results: The age of participants ranged from 24 to 68 years with mean age of 38.17 years and mean parity of 2.25. visual inspection of cervix with acetic acid and colposcopy were positive in eight (20%) and ten (25%) respectively. There were five (12.5%) cases of histopathologically proven lesion. The sensitivity of visual inspection of cervix with acetic acid and colposcopy were 80% and 100 % respectively and that of specificity were 88.5% and 85.5%.Conclusions: visual inspection of cervix with acetic acid is an effective screening tool with comparable sensitivity and specificity. It can be used as alternative screening methods especially in low income resource countries where the burden of disease is high.Keywords: Cervical cancer; Colposcopy; screening; VIA.
Sequence analysis has been widely used to investigate the patterns of similarities and differences of sequential data in biology and sociology. However, the debate on the usage of sequence analysis in social sciences has not been settled yet. Among a long list, sequence analysis methods have been criticized for ignoring the qualitative information behind the sequences. This paper presents a new instrument for inspecting sequential data visually in qualitative studies. The method includes building a hierarchical tree of relations among the categories which is then used to recode the categories systematically. The recoding process is meant to give meaning to the differences among categories and, therefore, increases our ability to see the differences. The instrument is a fruit of a qualitative study carried out to explore student's learning patterns. The focus in this paper will be on the algorithm of recoding the categories and how the emergent codes can be plotted to generate insights for further qualitative investigation.
AbstractOur Visual Analytics (VA) tool ScrutinAI supports human analysts to investigate interactively model performance and data sets. Model performance depends on labeling quality to a large extent. In particular in medical settings, generation of high quality labels requires in depth expert knowledge and is very costly. Often, data sets are labeled by collecting opinions of groups of experts. We use our VA tool to analyze the influence of label variations between different experts on the model performance. ScrutinAI facilitates to perform a root cause analysis that distinguishes weaknesses of deep neural network (DNN) models caused by varying or missing labeling quality from true weaknesses. We scrutinize the overall detection of intracranial hemorrhages and the more subtle differentiation between subtypes in a publicly available data set.
