| 田志新,廖 薇,茅 健,吴建民,袁 泉,徐 震.融合边缘监督的改进 Deeplabv3+水下鱼类分割方法[J].电子测量与仪器学报,2022,36(10):208-216 |
| 融合边缘监督的改进 Deeplabv3+水下鱼类分割方法 |
| Improved Deeplabv3+ underwater fish segmentationmethod combining with edge supervision |
| |
| DOI： |
| 中文关键词: 鱼类分割 边缘监督 Deeplabv3+ CBAM 注意力机制 混合膨胀卷积 |
| 英文关键词:fish segmentation edge supervision Deeplabv3+ CBAM attention mechanism hybrid dilated convolution |
| 基金项目:国家农业环境奉贤观测实验站项目（NAES035AE03）、上海市科技兴农项目（2022 02 08 00 12 F01186）、国家自然科学基金青年基金项目（62001282）资助 |
|
|
|
|
| 摘要点击次数: 822 |
| 全文下载次数: 741 |
| 中文摘要: |
| 水下环境鱼类分割是实现体长测量、体重估算和群体计数等智能化测量的关键技术,为了提高分割的准确性,提出一种
融合边缘监督的改进 Deeplabv3+鱼类分割方法。 编码部分采用更少的下采样次数,浅层增加卷积块注意力机制( convolutional
block attention module,CBAM),以减少信息丢失并增强浅层语义信息;通过设计混合膨胀卷积( hybrid dilated convolution,HDC)
改进空洞空间卷积池化金字塔(atrous spatial pyramid pooling,ASPP)模块,提取深层特征;在解码输出部分结合 Canny 边缘检测
算子引入边缘监督,通过边缘损失函数来获得边缘预测和边缘标签的误差以更好地学习边缘特征;最后根据不同类像素比率引
入优化的损失函数,进一步提升模型的语义分割性能。 该方法在 VOC2012 数据集上 mIoU 达到 84. 56%,较 Deeplabv3+方法提
升了 3. 27%,验证了其泛化能力。 在 DeepFish 数据集上做消融实验,mIoU 高达 93. 66%,均高于 Deeplabv3+、Unet 和 PSPNet 等
常见方法。 该研究提升了水下环境鱼类分割的精度,可为水产养殖智能化提供支持。 |
| 英文摘要: |
| Fish segmentation in underwater environment is the key technology to realize intelligent measurement such as body length
measurement, weight estimation and population counting. In order to improve the accuracy of fish segmentation, an improved Deeplabv3
+ fish segmentation method combined with edge supervision is proposed. In the encoder part, fewer down sampling times are used, and
convolutional block attention module (CBAM) is added in the shallow layer to reduce information loss and enhance the shallow semantic
information; By designing hybrid dilated convolution (HDC) to improve atrous spatial pyramid pooling(ASPP) module, deep features
are extracted. In the decoder output part, Canny edge detection operator is combined to introduce edge supervision, and the edge
prediction and edge label errors are obtained through the edge loss function to better learn edge features. Finally, the optimized loss
function is introduced according to different pixel ratios to further improve the semantic segmentation performance of the model. This
method achieves 84. 56% mIoU on VOC2012 dataset, which is 3. 27% higher than Deeplabv3+ method, and verifies its generalization
ability. In the ablation experiment on DeepFish dataset, mIoU is as high as 93. 66%, which is higher than common methods such as
Deeplabv3+, Unet and PSPNet. This research improves the accuracy of fish segmentation in underwater environment and can provide
support for intelligent aquaculture. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
