Spatially Attentive Output Layer for Image Classification

Kim, Ildoo; Baek, Woonhyuk; Kim, Sungwoong

doi:10.1109/cvpr42600.2020.00955

Cited by 25 publications

(12 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The convolutional neural network (CNN) has proven to be an effective computational model for automatically extracting image features [27][28][29]. A morphologicalbased convolutional neural network is constructed to grasp more morphological priors, which then guides the network to categorize the input wear particles images.…”

Section: Morphological Residual Convolutional Neural Network (M-rcnn)mentioning

confidence: 99%

Morphological residual convolutional neural network (M-RCNN) for intelligent recognition of wear particles from artificial joints

Song

Liao

et al. 2021

Friction

View full text Add to dashboard Cite

Finding the correct category of wear particles is important to understand the tribological behavior. However, manual identification is tedious and time-consuming. We here propose an automatic morphological residual convolutional neural network (M-RCNN), exploiting the residual knowledge and morphological priors between various particle types. We also employ data augmentation to prevent performance deterioration caused by the extremely imbalanced problem of class distribution. Experimental results indicate that our morphological priors are distinguishable and beneficial to largely boosting overall performance. M-RCNN demonstrates a much higher accuracy (0.940) than the deep residual network (0.845) and support vector machine (0.821). This work provides an effective solution for automatically identifying wear particles and can be a powerful tool to further analyze the failure mechanisms of artificial joints.

show abstract

Section: Morphological Residual Convolutional Neural Network (M-rcnn)mentioning

confidence: 99%

Morphological residual convolutional neural network (M-RCNN) for intelligent recognition of wear particles from artificial joints

Song

Liao

et al. 2021

Friction

View full text Add to dashboard Cite

show abstract

“…Recently, related work [18,19,26] has shown that teacher and student models can come from the same CNN network, and dynamically transfer knowledge by adding Side classifiers behind some intermediate layers, which is called self-knowledge distillation (SKD). In [27], Lee et al pointed out that adding auxiliary (Side) classifiers allows the intermediate layer to obtain gradient flows from both the topmost and branch losses, alleviating the "gradient disappearance" problem that occurs in the back propagation of gradients caused by deeper networks, and accelerating the convergence.…”

Section: Self-knowledge Distillationmentioning

confidence: 99%

Automatic Diabetic Retinopathy Grading via Self-Knowledge Distillation

2020

View full text Add to dashboard Cite

Diabetic retinopathy (DR) is a common fundus disease that leads to irreversible blindness, which plagues the working-age population. Automatic medical imaging diagnosis provides a non-invasive method to assist ophthalmologists in timely screening of suspected DR cases, which prevents its further deterioration. However, the state-of-the-art deep-learning-based methods generally have a large amount of model parameters, which makes large-scale clinical deployment a time-consuming task. Moreover, the severity of DR is associated with lesions, and it is difficult for the model to focus on these regions. In this paper, we propose a novel deep-learning technique for grading DR with only image-level supervision. Specifically, we first customize the model with the help of self-knowledge distillation to achieve a trade-off between model performance and time complexity. Secondly, CAM-Attention is used to allow the network to focus on discriminative zone, e.g., microaneurysms, soft/hard exudates, etc.. Considering that directly attaching a classifier after the Side branch will disrupt the hierarchical nature of convolutional neural networks, a Mimicking Module is employed that allows the Side branch to actively mimic the main branch structure. Extensive experiments are conducted on two benchmark datasets, with an AUC of 0.965 and an accuracy of 92.9% for the Messidor dataset and 67.96% accuracy achieved for the challenging IDRID dataset, which demonstrates the superior performance of our proposed method.

show abstract

“…With the development of computer hardware and a great improvement in computing power, deep learning (DL) has been successfully applied in image edge extraction, image segmentation [14,15], image classification [16,17], object detection [18,19], tracking, etc. When compared with the traditional edge detecting methods, the deep neural network can learn stable image features with high distinguishing ability, different levels, and different scales that are beneficial for edge detecting.…”

Section: Introductionmentioning

confidence: 99%

Edge Detecting Method for Microscopic Image of Cotton Fiber Cross-Section Using RCF Deep Neural Network

He¹,

Wang²

2021

Information

View full text Add to dashboard Cite

Currently, analyzing the microscopic image of cotton fiber cross-section is the most accurate and effective way to measure its grade of maturity and then evaluate the quality of cotton samples. However, existing methods cannot extract the edge of the cross-section intact, which will affect the measurement accuracy of maturity grade. In this paper, a new edge detection algorithm that is based on the RCF convolutional neural network (CNN) is proposed. For the microscopic image dataset of the cotton fiber cross-section constructed in this paper, the original RCF was firstly used to extract the edge of the cotton fiber cross-section in the image. After analyzing the output images of RCF in each convolution stage, the following two conclusions are drawn: (1) the shallow layers contain a lot of important edge information of the cotton fiber cross-section; (2) because the size of the cotton fiber cross-section in the image is relatively small and the receptive field of the convolutional layer gradually increases with the deepening of the number of layers, the edge information detected by the deeper layers becomes increasingly coarse. In view of the above two points, the following improvements are proposed in this paper: (1) modify the network supervision model and loss calculation structure; (2) the dilated convolution in the deeper layers is removed; therefore, the receptive field in the deeper layers is reduced to adapt to the detection of small objects. The experimental results show that the proposed method can effectively improve the accuracy of edge extraction of cotton fiber cross-section.

show abstract

Spatially Attentive Output Layer for Image Classification

Cited by 25 publications

References 26 publications

Morphological residual convolutional neural network (M-RCNN) for intelligent recognition of wear particles from artificial joints

Morphological residual convolutional neural network (M-RCNN) for intelligent recognition of wear particles from artificial joints

Automatic Diabetic Retinopathy Grading via Self-Knowledge Distillation

Edge Detecting Method for Microscopic Image of Cotton Fiber Cross-Section Using RCF Deep Neural Network

Contact Info

Product

Resources

About