Gabor capsule network with preprocessing blocks for the recognition of complex images

Ayidzoe, Mighty Abra; Yu, Yongbin; Mensah, Patrick Kwabena; Cai, Jingye; Adu, Kwabena; Tang, Yifan

doi:10.1007/s00138-021-01221-6

Cited by 19 publications

(11 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For gastrointestinal illnesses and GI tract disorders, this capsule can generate novel 3D pictures. In [ 20 ], the author proposed a Gabor capsule network for classifying complex images of the Kvasir dataset. The model achieves an overall accuracy of 91.50%.…”

Section: Methodsmentioning

confidence: 99%

“…In DL, there is always scope for improvement and most of the researchers have not been able to achieve remarkable classification performance. Their methodologies and approaches suffer from various hindering factors because they overlook some inherent hurdles of DL models and medical image datasets [ 15 , 16 , 20 ]. A comparative analysis of state-of-the-art WCE classification models is depicted in Table 1 .…”

Section: Methodsmentioning

confidence: 99%

“…Additionally, there is efficient model training of the suggested one presented in Figure 1 . The size of training parameters is smaller in comparison with [ 18 , 20 , 21 ]. The Grad-CAM heat-map algorithm is used to depict the visual characteristics of GI tract ailments which highlight the features which lead to classification of an image sample.…”

Section: The Proposed Gi-detection Modelmentioning

confidence: 99%

See 2 more Smart Citations

A Robust Deep Model for Classification of Peptic Ulcer and Other Digestive Tract Disorders Using Endoscopic Images

et al. 2022

View full text Add to dashboard Cite

Accurate patient disease classification and detection through deep-learning (DL) models are increasingly contributing to the area of biomedical imaging. The most frequent gastrointestinal (GI) tract ailments are peptic ulcers and stomach cancer. Conventional endoscopy is a painful and hectic procedure for the patient while Wireless Capsule Endoscopy (WCE) is a useful technology for diagnosing GI problems and doing painless gut imaging. However, there is still a challenge to investigate thousands of images captured during the WCE procedure accurately and efficiently because existing deep models are not scored with significant accuracy on WCE image analysis. So, to prevent emergency conditions among patients, we need an efficient and accurate DL model for real-time analysis. In this study, we propose a reliable and efficient approach for classifying GI tract abnormalities using WCE images by applying a deep Convolutional Neural Network (CNN). For this purpose, we propose a custom CNN architecture named GI Disease-Detection Network (GIDD-Net) that is designed from scratch with relatively few parameters to detect GI tract disorders more accurately and efficiently at a low computational cost. Moreover, our model successfully distinguishes GI disorders by visualizing class activation patterns in the stomach bowls as a heat map. The Kvasir-Capsule image dataset has a significant class imbalance problem, we exploited a synthetic oversampling technique BORDERLINE SMOTE (BL-SMOTE) to evenly distribute the image among the classes to prevent the problem of class imbalance. The proposed model is evaluated against various metrics and achieved the following values for evaluation metrics: 98.9%, 99.8%, 98.9%, 98.9%, 98.8%, and 0.0474 for accuracy, AUC, F1-score, precision, recall, and loss, respectively. From the simulation results, it is noted that the proposed model outperforms other state-of-the-art models in all the evaluation metrics.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: The Proposed Gi-detection Modelmentioning

confidence: 99%

See 1 more Smart Citation

A Robust Deep Model for Classification of Peptic Ulcer and Other Digestive Tract Disorders Using Endoscopic Images

et al. 2022

View full text Add to dashboard Cite

show abstract

“…By updating the dynamic routing mechanism between the master capsule and the digital capsule, high-level entity representation is obtained, which not only reduces network parameters but also avoids over-fitting. Through the experimental verification of MNIST datasets, compared with CNN, CapsNet has higher classification accuracy in digital recognition, traffic sign recognition, and medical image analysis [ 32 , 33 , 34 ].…”

Section: Proposed Methodsmentioning

confidence: 99%

Loop Closure Detection Based on Residual Network and Capsule Network for Mobile Robot

Zhang

Zheng

Tan

et al. 2022

Sensors

View full text Add to dashboard Cite

Loop closure detection based on a residual network (ResNet) and a capsule network (CapsNet) is proposed to address the problems of low accuracy and poor robustness for mobile robot simultaneous localization and mapping (SLAM) in complex scenes. First, the residual network of a feature coding strategy is introduced to extract the shallow geometric features and deep semantic features of images, reduce the amount of image noise information, accelerate the convergence speed of the model, and solve the problems of gradient disappearance and network degradation of deep neural networks. Then, the dynamic routing mechanism of the capsule network is optimized through the entropy peak density, and a vector is used to represent the spatial position relationship between features, which can improve the ability of image feature extraction and expression to optimize the overall performance of networks. Finally, the optimized residual network and capsule network are fused to retain the differences and correlations between features, and the global feature descriptors and feature vectors are combined to calculate the similarity of image features for loop closure detection. The experimental results show that the proposed method can achieve loop closure detection for mobile robots in complex scenes, such as view changes, illumination changes, and dynamic objects, and improve the accuracy and robustness of mobile robot SLAM.

show abstract

“…Then, we compare the test accuracy of our proposed model with the state-of-the-art based CapsNet models (Table 6) on the same datasets, and we analyse the number of training parameters and epochs (Table 7) on the CIFAR-10 dataset. [24] 70.33 --Max-min [25] 75.92 --Fsc-CapsNet [26] 78.73 --Residual CapsNet [27] 75.62 --STAR-Caps [28] 91.23 67.66 -RS-CapsNet [13] 89.81 64.14 96.50 Residual Capsule Network [14] 84.16 --HitNet [29] 73.30 -94.50 Quaternion CapsNet [30] 82.21 -95.37 Multi-lane [31] 76.79 --MS-CapsNet [32] 75.70 --Gabor CapsNet [33] 85.24 68.17 -DenseCaps [8] 89.41 -95.99 DA-CapsNet [34] 85 [6] 101.5 M 500 Multi-lane [31] 14.25 M 20 MS-CapsNet [32] 11.2 M 50 DenseCaps [8] 16 M 100 HitNet [29] 8.89M 250 Residual Capsule Network [14] 11.86M 120 ER-Caps 9.4 M 70 ER-Caps needs fewer epochs to converge (Fig. 7) than the baseline CapsNet, except on the CIFAR-100 dataset.…”

Section: Performance Evaluationmentioning

confidence: 99%

ER-Caps: ELU Residual Capsule Network for Complex Images Classification

2023

IJIES

View full text Add to dashboard Cite

Capsule Network has proved its performance against image deformation, through its ability to preserve space relations among features. However, the Capsule Network underperforms at classifying complex images datasets. Through an in-depth examination of Residual Neural Network, we present a novel Capsule Network named ER-caps to improve the model performance toward complex images classification and reconstruction. The architecture of ERcaps uses customized residual blocks to enhance feature extraction without using any pooling methods to conserve spatial information among features and preserve the Capsule Network equivariance. Our model adopts the RELU and the ELU activation function to ensure fast convergence and uses an advanced decoder to allow better reconstruction for complex images. The experiments show that the suggested model outperforms the baseline Capsule Network with dynamic routing on a variety of benchmark datasets CIFAR-10, CIFAR-100, and SVHN by approximately: 17 %, 9 %, and 4 % respectively. Moreover, ER-caps reduces the number of training parameters.

show abstract

Gabor capsule network with preprocessing blocks for the recognition of complex images

Cited by 19 publications

References 19 publications

A Robust Deep Model for Classification of Peptic Ulcer and Other Digestive Tract Disorders Using Endoscopic Images

A Robust Deep Model for Classification of Peptic Ulcer and Other Digestive Tract Disorders Using Endoscopic Images

Loop Closure Detection Based on Residual Network and Capsule Network for Mobile Robot

ER-Caps: ELU Residual Capsule Network for Complex Images Classification

Contact Info

Product

Resources

About