WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern
Wild fish recognition is a fundamental problem of ocean ecology research and contributes to the understanding of biodiversity. Given the huge number of wild fish species and unrecognized category, the essence of the problem is an open set fine-grained recognition. Moreover, the unrestricted marine e...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10197176/ |
| _version_ | 1827362045523132416 |
|---|---|
| author | Xiaoya Zhang Baoxiang Huang Ge Chen Milena Radenkovic Guojia Hou |
| author_facet | Xiaoya Zhang Baoxiang Huang Ge Chen Milena Radenkovic Guojia Hou |
| author_sort | Xiaoya Zhang |
| collection | DOAJ |
| description | Wild fish recognition is a fundamental problem of ocean ecology research and contributes to the understanding of biodiversity. Given the huge number of wild fish species and unrecognized category, the essence of the problem is an open set fine-grained recognition. Moreover, the unrestricted marine environment makes the problem even more challenging. Deep learning has been demonstrated as a powerful paradigm in image classification tasks. In this article, the wild fish recognition deep neural network (termed WildFishNet) is proposed. Specifically, an open set fine-grained recognition neural network with a fused activation pattern is constructed to implement wild fish recognition. First, three different reciprocal inverted residual structural modules are combined by neural structure search to obtain the best feature extraction performance for fine-grained recognition; next, a new fusion activation pattern of softmax and openmax functions is designed to improve the recognition ability of open set. Then, the experiments are implemented on the WildFish dataset that consists of 54 459 unconstrained images, which includes 685 known classes and 1 open set unrecognized category. Finally, the experimental results are analyzed comprehensively to demonstrate the effectiveness of the proposed method. The in-depth study also shows that artificial intelligence can empower marine ecosystem research. |
| first_indexed | 2024-03-08T07:19:18Z |
| format | Article |
| id | doaj.art-54d70ac087b74e548f13cf844d3c3863 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-03-08T07:19:18Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-54d70ac087b74e548f13cf844d3c38632024-02-03T00:00:55ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352023-01-01167303731410.1109/JSTARS.2023.329970310197176WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation PatternXiaoya Zhang0https://orcid.org/0009-0002-1535-746XBaoxiang Huang1https://orcid.org/0000-0002-0380-419XGe Chen2https://orcid.org/0000-0003-4868-5179Milena Radenkovic3https://orcid.org/0000-0003-4000-6143Guojia Hou4https://orcid.org/0000-0001-6509-6259Department of Computer Science and Technology, Qingdao University, Qingdao, ChinaDepartment of Computer Science and Technology, Qingdao University, Qingdao, ChinaLaboratory for Regional Oceanography and Numerical Modeling, Qingdao National Laboratory for Marine Science and Technology, Qingdao, ChinaSchool of Computer Science and Information Technology, The University of Nottingham, Nottingham, U.K.Department of Computer Science and Technology, Qingdao University, Qingdao, ChinaWild fish recognition is a fundamental problem of ocean ecology research and contributes to the understanding of biodiversity. Given the huge number of wild fish species and unrecognized category, the essence of the problem is an open set fine-grained recognition. Moreover, the unrestricted marine environment makes the problem even more challenging. Deep learning has been demonstrated as a powerful paradigm in image classification tasks. In this article, the wild fish recognition deep neural network (termed WildFishNet) is proposed. Specifically, an open set fine-grained recognition neural network with a fused activation pattern is constructed to implement wild fish recognition. First, three different reciprocal inverted residual structural modules are combined by neural structure search to obtain the best feature extraction performance for fine-grained recognition; next, a new fusion activation pattern of softmax and openmax functions is designed to improve the recognition ability of open set. Then, the experiments are implemented on the WildFish dataset that consists of 54 459 unconstrained images, which includes 685 known classes and 1 open set unrecognized category. Finally, the experimental results are analyzed comprehensively to demonstrate the effectiveness of the proposed method. The in-depth study also shows that artificial intelligence can empower marine ecosystem research.https://ieeexplore.ieee.org/document/10197176/Deep neural networkfusion activation patternneural structure searchopen set fine-grained recognitionwild fish recognition |
| spellingShingle | Xiaoya Zhang Baoxiang Huang Ge Chen Milena Radenkovic Guojia Hou WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Deep neural network fusion activation pattern neural structure search open set fine-grained recognition wild fish recognition |
| title | WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern |
| title_full | WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern |
| title_fullStr | WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern |
| title_full_unstemmed | WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern |
| title_short | WildFishNet: Open Set Wild Fish Recognition Deep Neural Network With Fusion Activation Pattern |
| title_sort | wildfishnet open set wild fish recognition deep neural network with fusion activation pattern |
| topic | Deep neural network fusion activation pattern neural structure search open set fine-grained recognition wild fish recognition |
| url | https://ieeexplore.ieee.org/document/10197176/ |
| work_keys_str_mv | AT xiaoyazhang wildfishnetopensetwildfishrecognitiondeepneuralnetworkwithfusionactivationpattern AT baoxianghuang wildfishnetopensetwildfishrecognitiondeepneuralnetworkwithfusionactivationpattern AT gechen wildfishnetopensetwildfishrecognitiondeepneuralnetworkwithfusionactivationpattern AT milenaradenkovic wildfishnetopensetwildfishrecognitiondeepneuralnetworkwithfusionactivationpattern AT guojiahou wildfishnetopensetwildfishrecognitiondeepneuralnetworkwithfusionactivationpattern |