Predicting the Breed of Dogs and Cats with Fine-Tuned Keras Applications

Wang, I.-Hung; Mahardi,; Lee, Kuang-Chyi; Chang, Shinn‐Liang

doi:10.32604/iasc.2021.019020

Cited by 3 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Xception is also a CNN-Based model architecture that is made of depth-wise separable convolutions layers and 36 convolutional layers that are widely used in image classification tasks such as in earlier domestic animal breed classification works [6].…”

Section: Datasetmentioning

confidence: 99%

“…However, Xception has a different default image input size of 299x299 so the image will be rescaled to 299x299 [6], after that it is rescaled further according to each of their architecture requirement.…”

Section: Fig 2 Model Training Processmentioning

confidence: 99%

“…Convolutional Neural Network (CNN), have been used in various image-based classification tasks such as in identifying and classifying many different types of animals the way CNN classifies images is similar to the neuron that is found in the brain with many nodes, layers, and activation function that ends with an output, the name Convolutional come from the fact that CNN has several convolutional layers that can automatically find an image feature from a process called training, there are many different types of CNN-Based model architecture such as VGG [3], EfficientNet [4] EfficientNetV2 [5], Xception [6].…”

Section: Introductionmentioning

confidence: 99%

“…There also has been a study for a fine-tuned pretrained model made to classify dog and cat breeds with multiple CNN-Based models with different architectures such as Inception-ResNet v2, ResNet152 v2, and Xception-based CDC, the result of the study shows that the Xception one having the best accuracy out of all of them with a training accuracy of 99.49%, testing accuracy 91.24% and validation accuracy of 99.21% [6].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Fusion of pretrained CNN models for cat breed classification: A comparative study

Hamdi,

Sunaryo,

Prasetyo

2023

E3S Web Conf.

View full text Add to dashboard Cite

Cat is one of the most popular domestic animals that human has domesticated for a long time, since then, there have been many breeds that can be difficult to identify with each breed having different health issues and care requirement, to resolve this problem we used Convolutional Neural Network (CNN) a widely used artificial intelligence deep learning model that has been used in many image classification problem, in this study we explored 11 different types of CNN-Based model architecture to be used in a fusion-based technique and fine-tune the model to further increase its performance, our results show that fusion model is a promising technique in classifying cat breeds that outperforms all of the individual CNN- Based model architecture with the 3 fusion model having an accuracy of 0.9053, precision of 0.9075, recall of 0.9053, and F1 score of 0.9016, additionally, fine-tuning only shows a small effect in increasing the fusion model performance.

show abstract

Section: Datasetmentioning

confidence: 99%

Section: Fig 2 Model Training Processmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Fusion of pretrained CNN models for cat breed classification: A comparative study

Hamdi,

Sunaryo,

Prasetyo

2023

E3S Web Conf.

View full text Add to dashboard Cite

show abstract

“…We use three highly effective and widely used architectures trained on the ImageNet Large-Scale Visual Recognition Challenge (ILSVRC), InceptionV3, Inception-ResNetV2, and Xception as base classifiers for the FNs (Byeon et al, 2020 ; Ali et al, 2021 ; Wang et al, 2021a , b ; Yildirim and Çinar, 2021 ). In theory, these three deep networks can be replaced with other networks based on specific classification tasks.…”

Section: Disease Detection Algorithm For Retinal Oct Based On An Fnmentioning

confidence: 99%

FN-OCT: Disease Detection Algorithm for Retinal Optical Coherence Tomography Based on a Fusion Network

Huang

Feng

et al. 2022

Front. Neuroinform.

View full text Add to dashboard Cite

Optical coherence tomography (OCT) is a new type of tomography that has experienced rapid development and potential in recent years. It is playing an increasingly important role in retinopathy diagnoses. At present, due to the uneven distributions of medical resources in various regions, the uneven proficiency levels of doctors in grassroots and remote areas, and the development needs of rare disease diagnosis and precision medicine, artificial intelligence technology based on deep learning can provide fast, accurate, and effective solutions for the recognition and diagnosis of retinal OCT images. To prevent vision damage and blindness caused by the delayed discovery of retinopathy, a fusion network (FN)-based retinal OCT classification algorithm (FN-OCT) is proposed in this paper to improve upon the adaptability and accuracy of traditional classification algorithms. The InceptionV3, Inception-ResNet, and Xception deep learning algorithms are used as base classifiers, a convolutional block attention mechanism (CBAM) is added after each base classifier, and three different fusion strategies are used to merge the prediction results of the base classifiers to output the final prediction results (choroidal neovascularization (CNV), diabetic macular oedema (DME), drusen, normal). The results show that in a classification problem involving the UCSD common retinal OCT dataset (108,312 OCT images from 4,686 patients), compared with that of the InceptionV3 network model, the prediction accuracy of FN-OCT is improved by 5.3% (accuracy = 98.7%, area under the curve (AUC) = 99.1%). The predictive accuracy and AUC achieved on an external dataset for the classification of retinal OCT diseases are 92 and 94.5%, respectively, and gradient-weighted class activation mapping (Grad-CAM) is used as a visualization tool to verify the effectiveness of the proposed FNs. This finding indicates that the developed fusion algorithm can significantly improve the performance of classifiers while providing a powerful tool and theoretical support for assisting with the diagnosis of retinal OCT.

show abstract