计算机应用 ›› 2021, Vol. 41 ›› Issue (8): 2358-2365.DOI: 10.11772/j.issn.1001-9081.2020101596

所属专题: 前沿与综合应用

• 前沿与综合应用 • 上一篇    下一篇

基于动态特征蒸馏的水工隧洞缺陷识别方法

黄继爽1,2, 张华1,2,3, 李永龙1,2,3, 赵皓1,2, 王皓冉3,4, 冯春成1,2,3   

  1. 1. 西南科技大学 信息工程学院, 四川 绵阳 621010;
    2. 特殊环境机器人技术四川省重点实验室(西南科技大学), 四川 绵阳 621010;
    3. 清华四川能源互联网研究院, 成都 610213;
    4. 水沙科学与水利水电工程国家重点实验室(清华大学), 北京 100084
  • 收稿日期:2020-10-14 修回日期:2020-12-07 出版日期:2021-08-10 发布日期:2021-01-27
  • 通讯作者: 黄继爽
  • 作者简介:黄继爽(1993-),男,四川绵阳人,硕士研究生,主要研究方向:深度学习、模型压缩;张华(1969-),男,四川绵阳人,教授,博士,主要研究方向:辐射环境机器人、模式识别、智能系统;李永龙(1983-),男,山西太原人,高级工程师,博士研究生,主要研究方向:模式识别、智能系统;赵皓(1991-),男,四川南充人,博士研究生,主要研究方向:模式识别、智能系统;王皓冉(1988-),男,河南南阳人,助理研究员,博士,主要研究方向:库坝安全、智能巡检;冯春成(1992-),男,四川达州人,博士研究生,主要研究方向:模式识别、智能系统。
  • 基金资助:
    国家重点研发计划项目(2019YFB1310503);国家“十三五”核能开发科研项目(20161295);四川省科技创新创业苗子工程(2020JDRC0130);中国大唐集团有限公司科学技术项目(CDT-TZK/SYD[2018]-010)。

Hydraulic tunnel defect recognition method based on dynamic feature distillation

HUANG Jishuang1,2, ZHANG Hua1,2,3, LI Yonglong1,2,3, ZHAO Hao1,2, WANG Haoran3,4, FENG Chuncheng1,2,3   

  1. 1. School of Information Engineering, Southwest University of Science and Technology, Mianyang Sichuan 621010, China;
    2. Robot Technology Used for Special Environment Key Laboratory of Sichuan Province(Southwest University of Science and Technology), Mianyang Sichuan 621010, China;
    3. Sichuan Energy Internet Research Institute under Tsinghua University, Chengdu Sichuan 610213, China;
    4. State Key Laboratory of Hydroscience and Engineering(Tsinghua University), Beijing 100084, China
  • Received:2020-10-14 Revised:2020-12-07 Online:2021-08-10 Published:2021-01-27
  • Supported by:
    This work is partially supported by the National Key Research and Development Program of China (2019YFB1310503), the Nuclear Energy Development and Scientific Research Project of the National "13th Five-Year Plan" (20161295), the Sichuan Science and Technology Innovation and Entrepreneurship Seedling Project (2020JDRC0130), the Science and Technology Project of China Datang Corporation Limited (CDT-TZK/SYD[2018]-010).

摘要: 针对水工隧洞缺陷识别任务中现有深度卷积神经网络(DCNN)对缺陷图像特征提取能力不足、识别种类少、推理耗时长的问题,提出一种基于动态特征蒸馏的缺陷自主识别方法。首先,通过深度曲线估计网络对图像进行优化,从而改善低照度环境下的图像质量;其次,构建加入注意力机制的动态卷积模块取代传统静态卷积,并且把得到的动态特征用于训练教师网络以获得更好的模型特征提取能力;最后,在知识蒸馏框架中融合鉴别器结构,以构造一种动态特征蒸馏损失,并通过鉴别器将动态特征知识从教师网络传递到学生网络,从而在大幅减少模型推理时间的同时实现六类缺陷的高精度识别。在四川某水电站水工隧洞缺陷数据集上对该方法和原有残差网络进行对比实验,结果表明该方法可达到96.15%的识别准确率,其模型参数量和推理时间分别降低到原来的1/2和1/6。通过实验结果可知,将缺陷图像的动态特征蒸馏信息融合到识别网络中能够提高水工隧洞缺陷的识别效率。

关键词: 水工隧洞, 缺陷识别, 动态卷积, 知识蒸馏, 模型压缩

Abstract: Aiming at the problems that the existing Deep Convolutional Neural Network (DCNN) have insufficient defect image feature extraction ability, few recognition types and long reasoning time in hydraulic tunnel defect recognition tasks, an autonomous defect recognition method based on dynamic feature distillation was proposed. Firstly, the deep curve estimation network was used to optimize the image to improve the image quality in low illumination environment. Secondly, the dynamic convolution module with attention mechanism was constructed to replace the traditional static convolution, and the obtained dynamic features were used to train the teacher network to obtain better model feature extraction ability. Finally, a dynamic feature distillation loss was constructed by fusing the discriminator structure in the knowledge distillation framework, and the dynamic feature knowledge was transferred from the teacher network to the student network through the discriminator, so as to achieve the high-precision recognition of six types of defects while significantly reducing the model reasoning time. In the experiments, the proposed method was compared with the original residual network on a hydraulic tunnel defect dataset of a hydropower station in Sichuan Province. The results show that this method has the recognition accuracy reached 96.15%, and the model parameter amount and reasoning time reduced to 1/2 and 1/6 of the original ones respectively. It can be seen from the experimental results that fusing the dynamic feature distillation information of the defect image into the recognition network can improve the efficiency of hydraulic tunnel defect recognition.

Key words: hydraulic tunnel, defect recognition, dynamic convolution, knowledge distillation, model compression

中图分类号: