Structural diagram of the multiscale attention aggregation (MSAA) module.

<p>This module uses different sizes of convolution kernels, such as 3 × 3, 5 × 5, and 7 × 7, to perform multiscale analysis on feature maps, with small convolution kernels capturing fine details of small defects and large convolution kernels focusing on the overall shape of large defects.</...

Full description

Saved in:

Bibliographic Details
Main Author:	Haiyan Zhang (176919) (author)
Other Authors:	Zining Zhao (22057378) (author), Yilin Liu (1991839) (author), Jiange Liu (21011660) (author), Tingmei Ma (22057381) (author), Kexiao Wu (21011663) (author), Zhiwen Zhuang (22057384) (author), Jiajun Wang (512021) (author)
Published:	2025
Subjects:	Sociology Space Science Environmental Sciences not elsewhere classified Biological Sciences not elsewhere classified Information Systems not elsewhere classified uses different sizes steel surface defects steel quality inspection reducing missed detections real time according experimental results show ensuring steel quality effective technical support avoiding detection omissions achieve efficient transformation 65 %, 2 36 %, respectively steel defect detection div >< p significantly improving accuracy semantic segmentation models deeplabv3 +</ p 19 %, defect size defect conditions traditional methods size differences paper presents multiscale analysis model ’ misjudgements caused feature maps false positives ela module dimensional space dimensional features defective areas convolutional layers convolution kernels backbone network adjust thresholds adaptively initializes accurately distinguish accurate segmentation
Tags:	Add Tag No Tags, Be the first to tag this record!

Structural diagram of the multiscale attention aggregation (MSAA) module.

Similar Items