Bridging the Gap between Laboratory and Field Experiments in American Eel Detection Using Transfer Learning and Convolutional Neural Network

An automatic system that utilizes data analytics and machine learning to identify adult American eel in data obtained by imaging sonars is created in this study. Wavelet transform has been applied to de-noise the ARIS sonar data and a convolutional neural network model has been built to classify eels and non-eel objects. Because of the unbalanced amounts of data in laboratory and field experiments, a transfer learning strategy is implemented to fine-tune the convolutional neural network model so that it performs well for both the laboratory and field data. The proposed system can provide important information to develop mitigation strategies for safe passage of out-migrating eels at hydroelectric facilities.

[1]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[2]  Changshui Zhang,et al.  DeepFish: Accurate underwater live fish recognition with a deep architecture , 2016, Neurocomputing.

[3]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[4]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[5]  Qiang Huang,et al.  Underwater target classification using wavelet packets and neural networks , 2000, IEEE Trans. Neural Networks Learn. Syst..

[6]  Pierre Baldi,et al.  Understanding Dropout , 2013, NIPS.

[7]  J. Geist,et al.  Comparison of sonar-, camera- and net-based methods in detecting riverine fish-movement patterns , 2018 .

[8]  Wang Heng,et al.  Transfer Learning with Convolutional Neural Networks for SAR Ship Recognition , 2018 .

[9]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[10]  Junyu Dong,et al.  Transferring deep knowledge for object recognition in Low-quality underwater videos , 2018, Neurocomputing.

[11]  Yuri V. Makarov,et al.  Standardized Software for Wind Load Forecast Error Analyses and Predictions Based on Wavelet-ARIMA Models -- Applications at Multiple Geographically Distributed Wind Farms , 2013, 2013 46th Hawaii International Conference on System Sciences.

[12]  L. Atallah,et al.  Using wavelet analysis to classify and segment sonar signals scattered from underwater sea beds , 2003 .

[13]  Qiang Yang,et al.  A Survey on Transfer Learning , 2010, IEEE Transactions on Knowledge and Data Engineering.

[14]  T. Mulligan,et al.  Classifying Sonar Images: Can a Computer-Driven Process Identify Eels? , 2008 .

[15]  Ludwig Bothmann,et al.  Realtime classification of fish in underwater sonar videos , 2016 .

[16]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.