Abstract
Morphology and hydraulic condition of rivers explain habitat and actual species distribution of fish species. However, detailed ground surveys to obtain such habitat information are generally complicated and costly. Therefore, we aimed to explore the possibility of integrating aerial photographs and image recognition technique as a supplementing approach for field surveys. For this purpose, we focused on one benthic species (Acanthogobius flavimanus) and two migratory species (Nipponocypris temminckii and Plecoglossus altivelis) as representative species. Their distribution in the Kanto region of Japan was obtained from the national census on river environments-riparian zone, while aerial photographs of the corresponding river sections were collected from Geographical Survey Institute of Japan. Then, convolutional neural network (CNN) was applied to model the fish distribution based on the riverine photographs. As per the results from hypothesis tests, CNN was capable of learning relevant attributes of the river channel appearance for distribution of A. flavimanus. The model performance was significantly correlated with the number of the training data for this species. For this species, the relatively dark water surface and wide channel width seem to be possible key factors. At the same time, for the other two species, this modeling approach was not as successful as A. flavimanus, while the model performance did not show significant differences among the three species. Although practical application of this approach is still challenging in terms of available data and model validity, it is worthwhile further exploring its applicability to other riverine species and regions.
This is a preview of subscription content, log in via an institution to check access.
References
Albawi, S., Mohammed, T. A., & Al-Zawi, S. (2018). Understanding of a convolutional neural network. Proceedings of 2017 International Conference on Engineering and Technology, ICET 2017, 2018-January.
Ando M et al (2019) Discrimination of camera trap images by deep learning. Nature 29(7553):1–73
Araújo FG, Williams WP, Bailey RG (2000) Fish assemblages as indicators of water quality in the middle Thames estuary, England (1980–1989). Estuaries 23(3):305–317
Buckland M, Gey F (1994) The relationship between recall and precision. J Am Soc Inform Sci 45(1):12–19
Camana M, Dala-Corte RB, Becker FG (2016) Relation between species richness and stream slope in riffle fish assemblages is dependent on spatial scale. Environ Biol Fishes 99(8–9):603–612
Chen, L., Chen, J., Hajimirsadeghi, H., & Mori, G. (2020). Adapting grad-CAM for embedding networks. Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020, 2783–2792.
Cho, J., Lee, K., Shin, E., Choy, G., & Do, S. (2015). How much data is needed to train a medical image deep learning system to achieve necessary high accuracy?
Elango, S. Ramachandran, N. Low, R. (2022). Autonomous mosquito habitat detection using satellite imagery and convolutional neural networks for disease risk mapping. Environmental Science, ArXiv.
Favorskaya M, Pakhirka A (2019) Animal species recognition in the wildlife based on muzzle and shape features using joint CNN. Procedia Comput Sci 159:933–942
Fawcett T (2006) An introduction to ROC analysis. Pattern Recogn Lett 27(8):861–874
Fialho AP, Oliveira LG, Tejerina-Garro FL, De Mérona B (2008) Fish-habitat relationship in a tropical river under anthropogenic influences. Hydrobiologia 598(1):315–324
Gomez Villa A, Salazar A, Vargas F (2017) Towards automatic wild animal monitoring: Identification of animal species in camera-trap images using very deep convolutional neural networks. Eco Inform 41:24–32
Guisan A, Thuiller W (2005) Predicting species distribution: offering more than simple habitat models. Ecol Lett 8(9):993–1009
Hand DJ (2009) Measuring classifier performance: a coherent alternative to the area under the ROC curve. Mach Learn 77(1):103–123
Harrison LR, Legleiter CJ, Overstreet BT, Bell TW, Hannon J (2020) Assessing the potential for spectrally based remote sensing of salmon spawning locations. River Res Appl 36(8):1618–1632
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2016-Decem, 770–778.
Huang J, Huang L, Wu Z, Mo Y, Zou Q, Wu N, Chen Z (2019) Correlation of fish assemblages with habitat and environmental variables in a headwater stream section of Lijiang River. China Sustainability (switzerland) 11(4):1–14
Kameyama S, Fukushima M, Han M, Kaneko M (2007) Spatio-temporal changes in habitat potential of endangered freshwater fish in Japan. Eco Inform 2(4):318–327
Kawanabe, K., Mizuno, N. (2001), Fresfwater fushes of Japan, Yama-kei Publishers.
Kobayashi A, Nakano H, Murakami T (2013) Measurements of stream velocity of Katsura-gawa River using the electromagnetic velocity meter. Kyoto Univ Educ Environ Educ Res Inst 21:75–82
Krizhevsky BA, Sutskever I, Hinton GE (2012) ImageNet classification with deep convolutional neural networks. Commun ACM 60(6):84–90
Mohanty SP, Hughes DP, Salathé M (2016) Using deep learning for image-based plant disease detection. Front Plant Sci 7:1–10
Neill WH (1979) Mechanisms of fish distribution in heterothermal environments. Integr Comp Biol 19(1):305–317
Pankhurst RJ (1979) Biological identification methods. Nature 278:673–674
Rauf HT, Lali MIU, Zahoor S, Shah SZH, Rehman AU, Bukhari SAC (2019) Visual features based automated identification of fish species using deep convolutional neural networks. Comput Electron Agric 167:105075
Tummers JS, Hudson S, Lucas MC (2016) Evaluating the effectiveness of restoring longitudinal connectivity for stream fish communities: towards a more holistic approach. Sci Total Environ 569–570:850–860
Veit, A., Wilber, M., & Belongie, S. (2016). Residual networks behave like ensembles of relatively shallow networks. Advances in Neural Information Processing Systems, 550–558.
Wang L, Seelbach PW, Hughes RM (2006) Introduction to landscape influences on stream habitats and biological assemblages. Am Fish Soc Symp 2006(48):1–23
Yoshimura C, Tockner K, Furumai H, Omura T (2005) Present state of rivers and streams in Japan. River Res Appl 21(2–3):93–112
Zeni JO, Casatti L (2014) The influence of habitat homogenization on the trophic structure of fish fauna in tropical streams. Hydrobiologia 726(1):259–270
Acknowledgements
This research was financially supported by School of Environment and Society, Tokyo Institute of Technology.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
See Figs.
Distribution of each specific species in the target river
5,
Aerial photographs showing river sections that were estimated as habitat of yellowfin goby with relatively high probability (0.61 < p < 0.95) on average by the CNN-based model based on the number of images of 60, 80, and 100. Panel A-F are in the same scale
6,
Aerial photographs showing river sections that ware estimated to be habitats of the dark chub with relatively high probability (0.5 < p < 0.71) on average by the CNN-based models based on the numbers of image of 60, 80, and 100. Panel A-D are in the same scale except for the panel E
7,
Aerial photographs showing river sections that ware estimated to be habitats of sweetfish with relatively high probability (0.84 < p < 1) on average by the CNN-based models based on the numbers of image of 60, 80, and 100. Panel A-F are in the same scale
8.
Rights and permissions
About this article
Cite this article
Nagata, S., Yoshimura, C., Ly, S. et al. Direct application of residual neural network to riverine aerial photography for estimating fish distribution. Landscape Ecol Eng 19, 687–698 (2023). https://doi.org/10.1007/s11355-023-00566-6
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11355-023-00566-6