Grad-Cam Visualization Of Arabic Letter Character Prediction DOI

Asroni Asroni,

Cahya Damarjati,

Dhimas Rizki Akbar

et al.

Published: Dec. 1, 2023

Arabic letters, commonly called hijaiyah present a considerable challenge in acquisition and mastery. Introducing letters is significant subject due to the inherent challenges associated with their composition. This study aims compare class activation visualization characters by employing custom model contrasting it widely used models, namely AlexNet LeNet. The employed utilizes Class Activation Mapping (CAM) technique demonstrate its understanding of character identification process effectively. approach facilitates observation key focal points when identifies certain character. identify elements that contribute effectiveness Convolutional Neural Network (CNN) accurately recognizing characters. will be achieved training CNN using substantial dataset specifically emphasizes recognition. employ visualize results. results this not only offer comprehensive comprehension model's detection. However, they also assist identifying any problems may arise during procedure. outcomes research would enhance capacity script, hence facilitating implementation assistance for handling text damaged or blurred. In investigation, was observed performance surpassed LeNet convolutional neural network models. Training on consisting 13,440 data points, notable accuracy rate 97.38%. Additionally, exhibited loss 9.07% at epoch 50. interim, demonstrated 96.15% 93.12%, losses 15.88% 21.90%.

Language: Английский

Neuro-XAI: Explainable deep learning framework based on deeplabV3+ and bayesian optimization for segmentation and classification of brain tumor in MRI scans DOI

Tallha Saeed,

Muhammad Attique Khan, Ameer Hamza

et al.

Journal of Neuroscience Methods, Journal Year: 2024, Volume and Issue: 410, P. 110247 - 110247

Published: Aug. 10, 2024

Language: Английский

Citations

11
Automatic mandibular third molar and mandibular canal relationship determination based on deep learning models for preoperative risk reduction DOI Creative Commons
Elham Tahsin Yasin, Mediha Erturk, Melek Taşsöker

et al.

Clinical Oral Investigations, Journal Year: 2025, Volume and Issue: 29(4)

Published: March 25, 2025

Language: Английский

Citations

0
DSC-YOLOv8n: An advanced automatic detection algorithm for urban flood levels DOI

Jiaquan Wan,

Yufang Shen, Fengchang Xue

et al.

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 643, P. 132028 - 132028

Published: Sept. 16, 2024

Language: Английский

Citations

3
Improvement of the YOLOv5 Model in the Optimization of the Brown Spot Disease Recognition Algorithm of Kidney Bean DOI Creative Commons

Pengyan Su,

Hao Li, Xiaoyun Wang

et al.

Plants, Journal Year: 2023, Volume and Issue: 12(21), P. 3765 - 3765

Published: Nov. 3, 2023

The kidney bean is an important cash crop whose growth and yield are severely affected by brown spot disease. Traditional target detection models cannot effectively screen out key features, resulting in model overfitting weak generalization ability. In this study, a Bi-Directional Feature Pyramid Network (BiFPN) Squeeze Excitation (SE) module were added to YOLOv5 improve the multi-scale feature fusion extraction abilities of improved model. results show that BiFPN SE modules higher heat location region pay less attention irrelevant environmental information non-target region. Precision, Recall, mean average Precision ([email protected]) 94.7%, 88.2%, 92.5%, respectively, which 4.9% 0.5% 25.6% compared original Compared with YOLOv5-SE, YOLOv5-BiFPN, FasterR-CNN, EfficientDet models, 1.8%, 3.0%, 9.4%, 9.5%, respectively. Moreover, rate missed wrong only 8.16%. Therefore, YOLOv5-SE-BiFPN can more detect area beans.

Language: Английский

Citations

8
WHEAT GRAINS AUTOMATIC COUNTING BASED ON LIGHTWEIGHT YOLOv8 DOI Open Access

Na Ma,

Zhongtao Li,

Qingzhong KONG

et al.

INMATEH Agricultural Engineering, Journal Year: 2024, Volume and Issue: unknown, P. 592 - 602

Published: Aug. 26, 2024

In order to accurately and quickly achieve wheat grain detection counting, efficiently evaluate quality yield, a lightweight YOLOv8 algorithm is proposed automatically count grains in different scenarios. Firstly, images are collected under three scenarios: no adhesion, slight severe create dataset. Then, the neck network of modified bidirectional weighted fusion BiFPN establish model. Finally, results counting statistically analyzed. Experimental show that after improvement with BiFPN, mAP (mean Average Precision) value 94.7%, reduction 12.3% GFLOPs. The improved model now requires only 9.34ms for inference occupies just 4.0MB memory. Compared other models, this paper performs best terms accuracy speed comprehensively, better meeting real-time requirements grains.

Language: Английский

Citations

2
Applications of Image Segmentation Techniques in Medical Images DOI Creative Commons
Yangyang Hou

ICST Transactions on e-Education and e-Learning, Journal Year: 2024, Volume and Issue: 10

Published: July 19, 2024

Image segmentation is an important research direction in medical image processing tasks, and it also a challenging task the field of computer vision. At present, there have been many methods, including traditional methods deep learning-based methods. Through understanding learning current situation segmentation, this paper systematically combs it. Firstly, briefly introduces such as threshold method, region method graph cut focuses on commonly used network architectures based CNN, FCN, U-Net, SegNet, PSPNet, Mask R-CNN. same time, application expounded. Finally, challenges development opportunities technology are discussed.

Language: Английский

Citations

1
Recent Progresses in Neural Networks for Alzheimer's Disease Detection DOI Open Access
Mengyao Zhao

Published: Oct. 12, 2023

This article reviews the introduction of Alzheimer's Disease (AD), neural networks, training and learning applications networks in early diagnosis AD, AD drug discovery, other brain diseases, challenges faced by AD. First, paper introduces background characteristics is a degenerative neurological disorder characterized impaired memory, decreased cognitive function, loss neurons. These place huge burden on lives families patients. Next, basic principle structure network are discussed. A computational model made up multiple neurons that can perform tasks adapting to input data. In particular, key concepts hierarchy, activation function weight adjustment Then, methods Common techniques such as backpropagation algorithm gradient descent optimizer introduced detail, well importance data preprocessing evaluation. focuses application By extracting features from image data, automatically identify differences between patients healthy subjects, enabling intervention. addition, discovery also analyzing predicting database known drugs, help discover potential treatments for speed process. The further explores diseases highlights lack reliable biomarkers, complex pathological mechanisms, etc. summary, this presents systematic overview associated with

Language: Английский

Citations

1
A K-Anonymous Location Privacy-Preserving Scheme for Mobile Terminals DOI Creative Commons
Weiping Peng, Di Ma, Cheng Song

et al.

ICST Transactions on e-Education and e-Learning, Journal Year: 2023, Volume and Issue: 9

Published: Dec. 11, 2023

Mobile terminals boost the prosperity of location-based service (LBS) which have already involved in every aspect People's daily life and are increasingly used various industries. Aimed at solving security efficiency problem existing location privacy protection schemes, a K-anonymity preservation scheme based on mobile terminal is proposed. Firstly, number rational dummy locations selected from cloaking region, more favorable further filtered according to entropy, so better anonymity effect can be achieved. Secondly, secure efficient m-out-of-n oblivious transfer protocol adopted, not only avoids dependency trusted center schemes improve efficiency, but also meets requirements for querying multiple interest points one time. Security analyses demonstrate that this satisfies such properties as anonymity, non-forgeability resistance replay attack, simulation results show has higher execution level, while low communications costs.

Language: Английский

Citations

1
Artificial Intelligence-based Mammogram Analysis for Early Detection DOI Open Access

Shengpei Ye

Published: Dec. 14, 2023

The article focuses on breast cancer, mammography, and artificial intelligence. First, cancer is a widespread health problem that affects millions of people worldwide, mammography widely adopted screening method. Then it introduced the advantages after AI participation, importance early detection application intelligence in treatment. From these aspects extend to entire medical field issues. Several times throughout paper, ethical issues could arise from applying healthcare are highlighted. At end article, paper describes continued development AI-based analysis over next period time.

Language: Английский

Citations

1
Machine Learning Methods for Handwriting Recognition DOI Open Access

Yibin Peng

Published: Dec. 18, 2023

Machine learning is a fundamental aspect of artificial intelligence that involves the development algorithms and models allow computers to learn make predictions or decisions without explicit programming. With neural networks, back-propagation deep learning, machine has made breakthroughs in fields image recognition, natural language processing handwriting recognition using techniques. The advent revolutionized field convolutional recurrent sequence-to-sequence provide solutions go beyond methods significantly improve accuracy robustness systems. But challenges remain, including need for large labelled datasets, computational resources addressing potential biases. As research techniques continues drive closer towards realisability, approaches remain at forefront.

Language: Английский

Citations

1