We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
融合通道注意力机制和非局部操作的蚕茧识别算法.
- Authors
叶飞; 汪小东; 王启真; 郭大容; 李子印; 杨娟亚
- Abstract
The cocoon cpiality has an important influence on the quality of silk, so it is necessary to classify the cocoons before reeling. In actual production, the traditional manual selecting method was high cost, low efficiency and accuracy* In order to improve the accuracy of silkworm cocoon classification deep learning method was used to improve silkworm cocoon classification based on ResNetSO. Firstly, ResNet50 was used, as the backbone network to extract the feature information of the Original cocoon image* Then y non-local operations were used to extract the global features of the feature image. Furthermore the channel attention mechanism was used to extract the channel features of the global feature image so that the global semantic features and details were pooled globally and the feature vector obtained from the global average pooling is input into the Softmax classifier for classification. The experimental results show that the average classification accuracy of the cocoon recognition algorithm combining the channel attention mechanism and non-loca] operation is 95. 6%, and. the average classification accuracy is 2.4% higher than that of the ResNetSO network mode.
- Subjects
COCOONS; SILKWORMS; DEEP learning; SILK; CLASSIFICATION; ALGORITHMS
- Publication
Wool Textile Journal, 2023, Vol 51, Issue 5, p115
- ISSN
1003-1456
- Publication type
Article
- DOI
10.19333/j.mfkj.20221104906