STDPNet: a dual-path surface defect detection neural network based on shearlet transform

Article Computer Science, Software Engineering

Split-guidance network for salient object detection

Shuhan Chen et al.

Summary: A simple yet efficient split-guidance convolution block is proposed to improve the multi-scale representation ability and a unified decoder for both RGB SOD and RGB-D SOD is built.

VISUAL COMPUTER (2023)

添加到收藏夹

Article Computer Science, Software Engineering

An improved hybrid multiscale fusion algorithm based on NSST for infrared-visible images

Peng Hu et al.

Summary: An improved hybrid multiscale fusion algorithm inspired by NSST is proposed to effectively extract and fuse complementary information in infrared-visible images, resulting in improved image contrast and retained saliency details.

VISUAL COMPUTER (2023)

添加到收藏夹

Article Computer Science, Software Engineering

RC-YOLOv5s: for tile surface defect detection

Wenqing Hou et al.

Summary: This paper proposes a deep learning-based detection model, RC-YOLOv5s, to solve the problems of complex surface texture of magnetic tile, cumbersome process of traditional detection algorithm, and low detection accuracy. The model incorporates two new structures: Res-Head and Drop-CA, which enhance feature fusion and information exchange, and reduce the missed detection rate. Compared with YOLOv5s, the proposed model improves the detection accuracy by 1.83%, reduces the missing rate from 1.673 to 0.372%, and achieves an average detection frame rate of 41.67, meeting the requirements of real-time and accuracy.

VISUAL COMPUTER (2023)

添加到收藏夹

Article Computer Science, Software Engineering

Global structure-guided learning framework for underwater image enhancement

Runjia Lin et al.

Summary: The proposed global-local-guided model effectively tackles color distortion and blurred edges in underwater image enhancement tasks through a coarse-to-fine approach. By integrating two neural networks, the model is able to restore clear textural details and vivid color in underwater images.

VISUAL COMPUTER (2022)

添加到收藏夹

Article Computer Science, Software Engineering

Online health status monitoring of high voltage insulators using deep learning model

Dipu Sarkar et al.

Summary: This paper proposes a system to monitor the health of ceramic insulators using aerial images and deep learning model, which combines YOLOv3 and Raspberry Pi for real-time operation and improved detection accuracy. Results demonstrate that the system is efficient in quickly and accurately identifying and classifying insulators.

VISUAL COMPUTER (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Real-time detector design for small targets based on bi-channel feature fusion mechanism

Xiuling Zhang et al.

Summary: YOLOv4-tiny is a simplified and fast detector, but it performs poorly in detecting small targets and occluded objects. BFF-YOLO, a real-time detector with bi-channel feature fusion, addresses this issue and achieves improved detection performance.

APPLIED INTELLIGENCE (2022)

添加到收藏夹

Article Computer Science, Software Engineering

Automatic detection of pulmonary nodules in CT images based on 3D Res-I network

Lukui Shi et al.

Summary: A model based on an improved 3D residual structure was proposed for the detection of pulmonary nodules. The model incorporates features like rectangular convolution kernel, grouping convolution, and pre-activation to enhance detection accuracy and reduce false positives and model size. Experimental results on the LUNA16 dataset show significant improvements in nodule detection.

VISUAL COMPUTER (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

A steel surface defect inspection approach towards smart industrial monitoring

Ruiyang Hao et al.

Summary: The study introduces a steel surface defect inspection model that extracts complex features from multi-shape steel surface defects using a deformable convolution enhanced backbone network. A feature fusion network with balanced feature pyramid generates high-quality multi-resolution feature maps for the inspection of multi-size defects, and a detector network achieves the localization and classification of steel surface defects. The proposed model achieves high efficiency in inference with a mAP of 0.805, 0.144 higher than baseline models, while maintaining a balance between inspection performance and inference efficiency.

JOURNAL OF INTELLIGENT MANUFACTURING (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Automatic fabric defect detection using a wide-and-light network

Jun Wu et al.

Summary: This paper proposes a wide-and-light network structure based on Faster R-CNN to improve the accuracy and reduce the computational cost of fabric defect detection. By designing dilated convolution modules and optimizing feature extraction networks, high-precision fabric defect detection is achieved.

APPLIED INTELLIGENCE (2021)

添加到收藏夹

Article Mathematical & Computational Biology