A YOLO-based Real-time Packaging Defect Detection System

Vu, Thi-Thu-Huyen; Pham, Dinh-Lam; Chang, Tai‐Woo

doi:10.1016/j.procs.2022.12.285

Cited by 24 publications

(13 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The degree of overlap between the bounding box of the raw data and the predicted bounding box is 0.5. The network loss function was defined by adding the error for object classification, the error for object reliability, and the error for the bounding box location and size, as shown in (5). The binary cross-entropy function was applied for object classification error and object reliability error, and the least squares error function was used for bounding box error.…”

Section: Experiments Results and Evaluation A Experiments Settingsmentioning

confidence: 99%

See 1 more Smart Citation

YOLO Network Optimization With a Single Circular Bounding Box for Detecting Defective Cigarettes

Park,

Park

2023

IEEE Access

View full text Add to dashboard Cite

The manufacturing industry utilizes computing technology, robot technology, artificial intelligence, and IoT to improve production processes and quality. In particular, object detection technology is used in various industrial fields, and object detection methods based on deep learning are attracting attention. The tobacco processing industry requires automated production facilities, and quality control for defects in product appearance is essential. Mainly because tobacco products are sold at high prices, poor appearance is a significant issue in terms of consumer complaints and processing costs. Therefore, accurate cigarette detection is essential. We propose a modified network structure based on the YOLOv4-Tiny network, and use it to build a network optimized for cigarette detection. The modified network uses a single circular bounding box for learning and fast detection. It utilizes visual techniques, such as gradient-weighted Class Activation Mapping (Grad-CAM) to analyze the degree of activation of the network to construct an optimal network. This reduces the size of the network and increases processing speed, while maintaining detection accuracy. This paper is expected to play an important role in quality control and efficient production in the manufacturing industry.INDEX TERMS Circular bounding box, detecting defective cigarettes, tobacco processing, You Only Look Once (YOLO).

show abstract

Section: Experiments Results and Evaluation A Experiments Settingsmentioning

confidence: 99%

“…It has been widely studied throughout academia and industry due to its diverse application potential. Object detection methods incorporating deep learning technology, which has excellent generalization and high classification performance while enabling real-time processing, are also widely used [5]- [7].…”

mentioning

confidence: 99%

YOLO Network Optimization With a Single Circular Bounding Box for Detecting Defective Cigarettes

Park,

Park

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The mean attained precision rate for test data was 92.34%. Some further applications of YOLO include road crack detection [ 32 ], semi-supervised YOLO for generic object detection [ 56 ], head detection [ 48 ], defect detection [ 28 ], YOLO-based a few-shot model [ 50 ], detecting small targets in infrared remote sensing [ 18 ], vehichle detection [ 5 ], traffic sign detection [ 52 ], colon cancer detection [ 24 ], and cattle body detection [ 31 ].…”

Section: Related Workmentioning

confidence: 99%

“… 56% [ 48 ] 2023 Wei et al HD-YOLO Thier approach was evaluated utilising PHDF, SEU-fisheye, HABBOF, and CEPDOF datasets. 98% [ 28 ] 2023 Vu et al YOLO In total, 400 photographs were taken: 200 of broken boxes and 200 of intact ones. 78.6% [ 50 ] 2023 Xia et al BC-YOLO PASCAL VOC 2007 and MS COCO 2014 datasets were used 43.9% [ 18 ] 2023 Li and Shen YOLOSR-IST Dataset from infrared image sequences (IRIS) and single-frame Infrared small target (SIRST) 99.2% [ 5 ] 2023 Bie et al YOLOv5n-L The BDD100K dataset was utilised.…”

Section: Related Workmentioning

confidence: 99%

Optimized deep learning vision system for human action recognition from drone images

Samma

Sama

2023

Multimed Tools Appl

View full text Add to dashboard Cite

There are several benefits to constructing a lightweight vision system that is implemented directly on limited hardware devices. Most deep learning-based computer vision systems, such as YOLO (You Only Look Once), use computationally expensive backbone feature extractor networks, such as ResNet and Inception network. To address the issue of network complexity, researchers created SqueezeNet, an alternative compressed and diminutive network. However, SqueezeNet was trained to recognize 1000 unique objects as a broad classification system. This work integrates a two-layer particle swarm optimizer (TLPSO) into YOLO to reduce the contribution of SqueezeNet convolutional filters that have contributed less to human action recognition. In short, this work introduces a lightweight vision system with an optimized SqueezeNet backbone feature extraction network. Secondly, it does so without sacrificing accuracy. This is because that the high-dimensional SqueezeNet convolutional filter selection is supported by the efficient TLPSO algorithm. The proposed vision system has been used to the recognition of human behaviors from drone-mounted camera images. This study focused on two separate motions, namely walking and running. As a consequence, a total of 300 pictures were taken at various places, angles, and weather conditions, with 100 shots capturing running and 200 images capturing walking. The TLPSO technique lowered SqueezeNet’s convolutional filters by 52%, resulting in a sevenfold boost in detection speed. With an F1 score of 94.65% and an inference time of 0.061 milliseconds, the suggested system beat earlier vision systems in terms of human recognition from drone-based photographs. In addition, the performance assessment of TLPSO in comparison to other related optimizers found that TLPSO had a better convergence curve and achieved a higher fitness value. In statistical comparisons, TLPSO surpassed PSO and RLMPSO by a wide margin.

show abstract

“…Shape and color defects are inevitable on flexible packaging prints due to equipment and environmental factors during the production processes, which include printing, laminating, * Authors to whom any correspondence should be addressed. curing, cutting and bag making [1]. The printing process is the most likely link to cause surface defects.…”

Section: Introductionmentioning

confidence: 99%

An online color and shape integrated detection method for flexible packaging surface defects

Sun,

Wei,

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

It is difficult for the spectrophotometer to meet the requirement of real-time color defect detection for flexible packaging prints. The false of shape defect detection is caused by artifact interference and insufficient classification accuracy of defect classification network. A color defect detection method for flexible packaging is proposed, which realizes the adaptive adjustment of the correction parameters of the Commission Internationale de l´Eclairage Delta E 2000 (CIEDE2000) equations with the detection object. It improves the speed and accuracy of the color defect detection for flexible packaging. An quadratic difference strategy is designed for template matching subtraction method to remove artifact interference. A method for enhancing shape defect data set of flexible packaging is proposed. Using discrete images of defects as network input, self-attention (SA) mechanism and spectral normalization (SN) methods are added to the deep convolutional generative adversarial networks (DCGAN) to enhance the effective dataset for the training of defect classification network. The accuracy of color detection for flexible packaging prints is improved by 38.7% based on optimized CIEDE2000 The average structure similarity index measure (SSIM) value of the improved DCGAN for defect detection is 0.845, and the Fréchet inception distance (FID) is 121.463. It takes 83.63 ms for the color and shape integrated detection method to detect shape defects on flexible packaging surfaces with an accuracy of 98.3%. The online color and shape integrated detection method can be applied to automated flexible packaging workshops to achieve real-time defect detection.

show abstract

A YOLO-based Real-time Packaging Defect Detection System

Cited by 24 publications

References 26 publications

YOLO Network Optimization With a Single Circular Bounding Box for Detecting Defective Cigarettes

YOLO Network Optimization With a Single Circular Bounding Box for Detecting Defective Cigarettes

Optimized deep learning vision system for human action recognition from drone images

An online color and shape integrated detection method for flexible packaging surface defects

Contact Info

Product

Resources

About