• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.8
• /
• pp.2199-2213
• /
• 2024

To enhance the quality of defect detection for Printed Circuit Board Assembly (PCBA) during electronic product manufacturing, this study primarily focuses on optimizing the YOLOv7-based method for PCBA defect detection. In this method, the Mish, a smoother function, replaces the Leaky ReLU activation function of YOLOv7, effectively expanding the network's information processing capabilities. Concurrently, a Squeeze-and-Excitation attention mechanism (SEAM) has been integrated into the head of the model, significantly augmenting the precision of small target defect detection. Additionally, considering angular loss, compared to the CIoU loss function in YOLOv7, the SIoU loss function in the paper enhances robustness and training speed and optimizes inference accuracy. In terms of data preprocessing, this study has devised a brightness adjustment data enhancement technique based on split-filtering to enrich the dataset while minimizing the impact of noise and lighting on images. The experimental results under identical training conditions demonstrate that our model exhibits a 9.9% increase in mAP value and an FPS increase to 164 compared to the YOLOv7. These indicate that the method proposed has a superior performance in PCBA defect detection and has a specific application value.

• Korean Journal of Construction Engineering and Management
• /
• v.24 no.5
• /
• pp.35-43
• /
• 2023

Along with the increase in Multi-unit housing defect disputes, the importance of defect management is also increased. However, previous studies have mostly focused on the Multi-unit housing's 'common part'. In addition, there is a lack of research on the system for the 'management office', which is a part of the subject of defect management. These resulted in the lack of defect management capability of the management office and the deterioration of management quality. Therefore, this paper proposes a machine learning-based defect data management system for management offices. The goal is to solve the inconvenience of management by using Optical Character Recognition (OCR) and Natural Language Processing (NLP) modules. This system converts handwritten defect information into online text via OCR. By using the language model, the defect information is regenerated along with the form specified by the user. Eventually, the generated text is stored in a database and statistical analysis is performed. Through this chain of system, management office is expected to improve its defect management capabilities and support decision-making.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.6
• /
• pp.1973-1979
• /
• 2000

Driving the need for machine vision system is growing consumer demand for quality and defect-free products. Especially it is the most important in tarpaulin's manufacturing process achieves automatically by machine vision instead of by man vision. In this paper pinholes detection is performed by using morphology algorithms. Top hat transform is one of morphology applications. This transform take high performance of defect detection in the case that unexpected changes occur in some non-uniform background. For pinholes defect, automatic visual inspection system has been developed, which was composed by a line-scan camera, illumination, a frame grabber, a motor driver and control units. This system has excellent capacity to defect pinholes to the 0.1 mm by 0.5 mm in size and to work in moving objects by maximum 20 m/min in speed.

• Proceedings of the IEEK Conference
• /
• 1999.11a
• /
• pp.637-640
• /
• 1999

In this Paper we show the LCD simulator for defect inspection using image processing algorithm and neural network. The defect inspection algorithm of the LCD consists of preprocessing, feature extraction and defect classification. Preprocess removes noise from LCD image, using morphology operator and neural network is used for the defect classification. Sample images with scratch, pinhole, and spot from real LCD color filter image are used. The proposed algorithms show that defect detected and classified in the ratio of 92.3% and 94.6 respectively.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.6
• /
• pp.369-376
• /
• 2013

Finance Information System is critical national infrastructure. Therefore it is important to select variables of defect causal factor for the system defect management effectively. We research and analyze detected errors in A Company's Finance Information System for three years. In the result of research and analysis, we have selected 9 variables of defect factor: the trading volume, the fluctuation of KOSDAQ index, and the number of public announcements, etc. Then we have assumed that these variables affect real system errors and analyzed correlation between the hypothesis and the detected system errors. After analyzing, we have extracted the trading volume, the number of orders and fills, changing tasks, and the fluctuations of NASDAQ index as valid variables of defect factor. These variables are proposed for failure prediction model as the variables to manage defects in the finance information system afterward.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2015.11a
• /
• pp.87-88
• /
• 2015

A defect dispute surrounding an apartment building is in constant trouble, a defect consulting is necessary that objective and impartial for solving a defect. The defect dispute about a defect lawsuit is conducted in the order that filling of lawsuit, consulting order, defect inspecting, estimating a repairing cost, writing a consulting report, submitting a consulting report, and decision by court. Of these, a step of defect inspecting is extensively investigated an occurred defect that each defect index and type from each part and place. At this time, it is collected of many data and created many information. For this, it need to organize and manage. The study is a modeling of field investigation work process that second phase of defect inspecting step. A literature study is defined a work until level 2. This study is defined the work until level 3 to 4. In addition, the modeling can do for using a job name, a place to job, a job to do, and a person concerned about defect consulting case. The modeling is expected a contribution of improving a defect consulting process and systematizing a judgment standard.

• Journal of information and communication convergence engineering
• /
• v.10 no.3
• /
• pp.295-299
• /
• 2012

Defect generation during organic bottom anti-reflective coating (BARC) in the photo lithography process is closely related to humidity control in the BARC coating unit. Defects are related to the water component due to the humidity and act as a blocking material for the etching process, resulting in an extreme pattern bridging in the subsequent BARC etching process of the poly etch step. In this paper, the lower limit for the humidity that should be stringently controlled for to prevent defect generation during BARC coating is proposed. Various images of defects are inspected using various inspection tools utilizing optical and electron beams. The mechanism for defect generation only in the specific BARC coating step is analyzed and explained. The BARC defect-induced gate pattern bridging mechanism in the lithography process is also well explained in this paper.

• The KIPS Transactions:PartD
• /
• v.10D no.4
• /
• pp.709-718
• /
• 2003

This research introduces defect and its causes that happen on software development. Based on defect cause analysis, we understand associated relation between defects and them design defect trigger. So, when we achieve similar project, we can forecast defect and prepare to solve defect by using defect trigger.

• Journal of Korea Society of Industrial Information Systems
• /
• v.19 no.6
• /
• pp.43-51
• /
• 2014

TFT-LCD image includes a defect of various properties. TFT-LCD image have a recognizable defects in the human inspector. On the other hand, it is difficult to detect defects that difference between the background and defect is very low. In this paper, we proposed sequentially detect algorithm from pixels included in the defect region to limited defects. And blob analysis methods using the blob size and gray difference are applied to the defect candidate image. Finally, we detect an accurate defect blob to distinguish the noise. The experimental results show that the proposed method finds the various defects reliably.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.108-111
• /
• 2022

Applying deep learning to machine vision systems for defect detection of products requires vast amounts of training data about various defect cases. However, since data imbalance occurs according to the type of defect in the actual manufacturing industry, it takes a lot of time to collect product images enough to generalize defect cases. In this paper, we apply a Siamese neural network that can be learned with even a small amount of data to product defect detection, and modify the image pairing method and contrastive loss function by properties the situation of product defect image data. We indirectly evaluated the embedding performance of Siamese neural networks using AUC-ROC, and it showed good performance when the images only paired among same products, not paired among defective products, and learned with exponential contrastive loss.