Reptile Search Algorithm with Deep Convolutional Neural Network for Cloud Assisted Colorectal Cancer Detection and Classification DOI Open Access
Shamik Tiwari

Tuijin Jishu/Journal of Propulsion Technology, Journal Year: 2023, Volume and Issue: 44(4), P. 1057 - 1073

Published: Oct. 16, 2023

Cloud-based automatic colorectal cancer (CC) detection involves the usage of cloud computing technology and system to help in earlier accurate diagnosis CC medical images patient information. This cloud-based aims improve efficiency reliability screening, monitoring, diagnoses. Automatic refers use computer-based systems aid data images. automated increase diagnosis. Deep learning (DL) methods, especially convolutional neural networks (CNNs), exhibit promising results They can be trained on wide-ranging datasets learn patterns features related precancerous cancerous lesion. study develops a new Reptile Search Algorithm with Learning for Colorectal Cancer Detection Classification (RSADL-CCDC) technique. The main aim RSADL-CCDC method focuses automaticclassification recognition environment. Once are stored server, process is carried out. In presented approach, initial stage preprocessing performed by bilateral filtering (BF) approach. For feature extraction, technique applies ShuffleNetv2 model. Besides, classification take place using autoencoder (CAE) Finally, hyperparameter tuning CAE takes utilizing RSA. experimental validation benchmark database. Extensive stated enhanced performance over other models respect tovarious actions.

Language: Английский

Steel Surface Defect Detection Technology Based on YOLOv8-MGVS DOI Creative Commons
Kai Zeng, Z. Cedric Xia, Junlei Qian

et al.

Metals, Journal Year: 2025, Volume and Issue: 15(2), P. 109 - 109

Published: Jan. 23, 2025

Surface defects have a serious detrimental effect on the quality of steel. To address problems low efficiency and poor accuracy in manual inspection process, intelligent detection technology based machine learning has been gradually applied to steel surface defects. An improved YOLOv8 defect model called YOLOv8-MGVS is designed these challenges. The MLCA mechanism C2f module increase feature extraction ability backbone network. lightweight GSConv VovGscsp cross-stage fusion modules are added neck network reduce loss semantic information achieve effective fusion. self-attention exploited into improve small targets. Defect experiments were carried out NEU-DET dataset. Compared with YOLOv8n from experimental results, average accuracy, recall rate, frames per second by 5.2%, 10.5%, 6.4%, respectively, while number parameters computational costs reduced 5.8% 14.8%, respectively. Furthermore, generalization GC-10 dataset SDD DET confirmed that higher better lightweight, speed.

Language: Английский

Citations

0
Defect measurement method of circular saw blade based on machine vision DOI
Hui Wang,

Yangyu Wang,

Pengcheng Ni

et al.

The International Journal of Advanced Manufacturing Technology, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 19, 2025

Language: Английский

Citations

0
A new network model for multiple object detection for autonomous vehicle detection in mining environment DOI Creative Commons

Muhammad Wahab Hanif,

Zhenhua Yu, Rehmat Bashir

et al.

IET Image Processing, Journal Year: 2024, Volume and Issue: 18(12), P. 3277 - 3287

Published: June 29, 2024

Abstract Considering the challenges of low multi‐object detection accuracy and difficulty in identifying small targets caused by challenging environmental conditions including irregular lighting patterns ambient noise levels mining environment with autonomous electric locomotives. A new network model based on SOD−YOLOv5s−4L has been proposed to detect multi‐objects for locomotives underground coal mines. Improvements have applied YOLOv5s construct model, introducing SIoU loss function address mismatch between real predicted bounding box directions, facilitating learn target position information more efficiently. This research introduces a decoupled head enhance feature fusion improve positioning precision enabling rapid capture multi‐scale features. Furthermore, capability increased layer which is developed increasing number layers from three four. The experimental results multiple object dataset show that achieves significant improvement mean average (mAP) almost 98% various types an (AP) nearly 99% other hand it 5.19% 9.79% compared model. comparative analysis models like YOLOv7 YOLOv8 shows superior performance terms detection.

Language: Английский

Citations

0
Liquid phase fluidity study of iron ore fines based on improved CondInst DOI
Meng Wang, Zhe Li,

Weixing Liu

et al.

Ironmaking & Steelmaking Processes Products and Applications, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 12, 2024

The liquid phase flowability of iron ore powder affects the quality sintered ore. To better explore flow law powder, this paper first adopted improved CondInst (Conditional Convolutions for Instance Segmentation) to segment image and achieved a segmentation accuracy 96.61%. accuracies on ResNet50 as well ResNet101 were by 0.11% 0.32%, respectively, relative original model. image's height, area wetting angle used characteristic indexes melting. fitting curve was established combining temperature time characterise whole process Second, Factsage simulate generation CatBoost regression model based constructed, maximum error between predicted value real 3.74%. Finally, equivalent mobility number, performance mechanism alkalinity's influence it comprehensively analysed.

Language: Английский

Citations

0
Reptile Search Algorithm with Deep Convolutional Neural Network for Cloud Assisted Colorectal Cancer Detection and Classification DOI Open Access
Shamik Tiwari

Tuijin Jishu/Journal of Propulsion Technology, Journal Year: 2023, Volume and Issue: 44(4), P. 1057 - 1073

Published: Oct. 16, 2023

Cloud-based automatic colorectal cancer (CC) detection involves the usage of cloud computing technology and system to help in earlier accurate diagnosis CC medical images patient information. This cloud-based aims improve efficiency reliability screening, monitoring, diagnoses. Automatic refers use computer-based systems aid data images. automated increase diagnosis. Deep learning (DL) methods, especially convolutional neural networks (CNNs), exhibit promising results They can be trained on wide-ranging datasets learn patterns features related precancerous cancerous lesion. study develops a new Reptile Search Algorithm with Learning for Colorectal Cancer Detection Classification (RSADL-CCDC) technique. The main aim RSADL-CCDC method focuses automaticclassification recognition environment. Once are stored server, process is carried out. In presented approach, initial stage preprocessing performed by bilateral filtering (BF) approach. For feature extraction, technique applies ShuffleNetv2 model. Besides, classification take place using autoencoder (CAE) Finally, hyperparameter tuning CAE takes utilizing RSA. experimental validation benchmark database. Extensive stated enhanced performance over other models respect tovarious actions.

Language: Английский

Citations

0