A Two-Stage Multiscale Residual Attention Network for Light Guide Plate Defect Detection

Li, Zhaopan; Li, Junfeng; Dai, Wei

doi:10.1109/access.2020.3047221

Cited by 20 publications

(18 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To explore the efficacy of the proposed CellDefectNet, we evaluated its performance on a benchmark dataset comprising of a diversity of photovoltaic cells captured using electroluminescence imagery [4]. The benchmark dataset comprises of 2624 images captured of monocrystalline and polycrystalline photovoltaic cells with 1100 defective samples and 1514 nondefective samples, with a training/test split of 75%/25% as described in [8]. The resolution of the images are 300 × 300.…”

Section: Results and Discussion A Experimental Setupmentioning

confidence: 99%

“…processing tasks [3], [18]. This even motivates researchers in the field of manufacturing to improve the automation including by developing different manufacturing tasks including the inspection systems [7], [8], [14]. However, the development in this this area is still in its infancy because several constraints and limitations need to be taken into account for these types of systems including, i) high efficiency and very fast runtime requirements, ii) high accuracy and robustness of the underlying machine learning model, and iii) the limitation on the number of available training data samples of different defected objects.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection

Xu¹,

Famouri²,

Bathla³

et al. 2022

Preprint

View full text Add to dashboard Cite

Photovoltaic cells are electronic devices that convert light energy to electricity, forming the backbone of solar energy harvesting systems. An essential step in the manufacturing process for photovoltaic cells is visual quality inspection using electroluminescence imaging to identify defects such as cracks, finger interruptions, and broken cells. A big challenge faced by industry in photovoltaic cell visual inspection is the fact that it is currently done manually by human inspectors, which is extremely time consuming, laborious, and prone to human error. While deep learning approaches holds great potential to automating this inspection, the hardware resource-constrained manufacturing scenario makes it challenging for deploying complex deep neural network architectures. In this work, we introduce CellDefectNet, a highly efficient attention condenser network designed via machine-driven design exploration specifically for electroluminesence-based photovoltaic cell defect detection on the edge. We demonstrate the efficacy of CellDefectNet on a benchmark dataset comprising of a diversity of photovoltaic cells captured using electroluminescence imagery, achieving an accuracy of ∼86.3% while possessing just 410K parameters (∼13× lower than EfficientNet-B0, respectively) and ∼115M FLOPs (∼12× lower than EfficientNet-B0) and ∼13× faster on an ARM Cortex A-72 embedded processor when compared to EfficientNet-B0.

show abstract

Section: Results and Discussion A Experimental Setupmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection

Xu¹,

Famouri²,

Bathla³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Although it has a high classification accuracy, it cannot complete defect positioning, which does not meet the needs of a new generation of vehicle light guide plate defect detection systems. Li et al [12] proposed a two-stage multiscale residual attention network based on 'segmentation + decision-making', using a U-shaped network structure to construct a segmented subnet, and designed a multiscale residual attention network for defect semantic segmentation units. A decision-making subnet was constructed to accurately determine whether the LGP image was defective.…”

Section: Introductionmentioning

confidence: 99%

Surface defect detection of vehicle light guide plates based on an improved RetinaNet

Li¹,

Wang²

2022

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Aiming at the vehicle navigation light guide plate (LGP) image characteristics, such as complex and gradient texures, uneven brightness, and small defects, this paper proposes a visual inspection method for LGP defects based on an improved RetinaNet. First, we use ResNeXt50 with higher accuracy under the same parameters as the backbone network, and propose the lightweight module Ghost_module to replace the 1×1 convolution in the lower half of the ResNeXt_block. This can reduce the resource parameters and consumption, and speed up training and inference. Second, we propose and use an improved feature pyramid network (IFPN) module to improve the feature fusion network in RetinaNet. It can more effectively fuse the shallow semantic information and high-level semantic information in the backbone feature extraction network, and further improve the detection ability of small target defects. Finally, the defect detection dataset constructed based on the vehicle LGP images collected at a industrial site, and experiments are performed on the vehicle LGP dataset and Aluminum Profile Defect Identification dataset (Aluminum Profile DID). The experimental results show that the proposed method is both efficient and effective. It achieves a better average detection rate of 98.6% on the vehicle LGP dataset. The accuracy and real-time performance can meet the requirements of industrial detection.

show abstract

“…The existing display defect detection methods are mainly divided into three types: methods based on image registration [ 2 , 3 , 4 ], background reconstruction [ 5 , 6 , 7 , 8 , 9 , 10 ], and deep learning [ 11 , 12 , 13 , 14 , 15 , 16 , 17 ]. Shuai et al [ 2 ] proposed the method of histogram equalization to adjust the brightness of the registered image, which can effectively suppress the problem of edge afterimages caused by the unaligned edges, and extract multiscale defects.…”

Section: Introductionmentioning

confidence: 99%

Crosstalk Defect Detection Method Based on Salient Color Channel Frequency Domain Filtering

Xie

Chen

Liu

et al. 2022

Sensors

View full text Add to dashboard Cite

Display crosstalk defect detection is an important link in the display quality inspection process. We propose a crosstalk defect detection method based on salient color channel frequency domain filtering. Firstly, the salient color channel in RGBY is selected by the maximum relative entropy criterion, and the color quaternion matrix of the displayed image is formed with the Lab color space. Secondly, the image color quaternion matrix is converted into the logarithmic spectrum in the frequency domain through the hyper-complex Fourier transform. Finally, Gaussian threshold band-pass filtering and hyper-complex inverse Fourier transform are used to separate the low-contrast defects and background of the display image. The experimental results show that the accuracy of the proposed algorithm reaches 96% for a variety of crosstalk defect detection. Compared with the current advanced defect detection algorithms, the effectiveness of the proposed method for low-contrast crosstalk defect detection is confirmed.

show abstract

A Two-Stage Multiscale Residual Attention Network for Light Guide Plate Defect Detection

Cited by 20 publications

References 24 publications

CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection

CellDefectNet: A Machine-designed Attention Condenser Network for Electroluminescence-based Photovoltaic Cell Defect Inspection

Surface defect detection of vehicle light guide plates based on an improved RetinaNet

Crosstalk Defect Detection Method Based on Salient Color Channel Frequency Domain Filtering

Contact Info

Product

Resources

About