Detecting aquaculture with deep learning in a low-data setting.

Detecting aquaculture with deep learning in a low-data setting.

Aquaculture is growing rapidly in the Amazon basin and detailed spatial information is needed to understand the trade-offs between food production, economic development, and environmental impacts. Large open-source datasets of medium resolution satellite imagery offer the potential for mapping a variety of infrastructure, including aquaculture ponds. However, there are many challenges utilizing this data, including few labelled examples, class imbalance, and spatial bias. We find previous rule-based methods for mapping aquaculture perform poorly in the Amazon. By incorporating temporal information through percentile data, we show deep learning models can outperform previous methods by as much as 15% with as few as 300 labelled examples. Further, generalization to unseen regions can be improved by incorporating segmentation information through masked pooling and using contrastive pretraining to harness large quantities of unlabelled data.

Saved in:
Bibliographic Details
Main Authors: GREENSTREET, L., FAN, J., PACHECO, F. S., BAI, Y., UMMUS, M. E., DORIA, C., BARROS, N. O., FORSBERG, B. R., XU, X., FLECKER, A., GOMES, C.
Other Authors: LAURA GREENSTREET, CORNELL UNIVERSITY; JOSHUA FAN, CORNELL UNIVERSITY; FELIPE SIQUEIRA PACHECO, CORNELL UNIVERSITY; YIWEI BAI, CORNELL UNIVERSITY; MARTA EICHEMBERGER UMMUS, CNPASA; CAROLINA DORIA, UNIVERSIDADE FEDERAL DE RONDÔNIA; NATHAN OLIVEIRA BARROS, UNIVERSIDADE FEDERAL DE JUIZ DE FORA; BRUCE R. FORSBERG, INPA; XIANGTAO XU, CORNELL UNIVERSITY; ALEXANDER FLECKER, CORNELL UNIVERSITY; CARLA GOMES, CORNELL UNIVERSITY.
Format: Artigo em anais e proceedings biblioteca
Language:Ingles
English
Published: 2024-01-25
Subjects:Image segmentation, Image classification, Attention, Contrastive learning, Representation learning, Convolutinal neural networks, Sensoriamento Remoto, Aquicultura, Aquaculture, Remote sensing, Digital images, Neural networks,
Online Access:http://www.alice.cnptia.embrapa.br/alice/handle/doc/1161305
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Aquaculture is growing rapidly in the Amazon basin and detailed spatial information is needed to understand the trade-offs between food production, economic development, and environmental impacts. Large open-source datasets of medium resolution satellite imagery offer the potential for mapping a variety of infrastructure, including aquaculture ponds. However, there are many challenges utilizing this data, including few labelled examples, class imbalance, and spatial bias. We find previous rule-based methods for mapping aquaculture perform poorly in the Amazon. By incorporating temporal information through percentile data, we show deep learning models can outperform previous methods by as much as 15% with as few as 300 labelled examples. Further, generalization to unseen regions can be improved by incorporating segmentation information through masked pooling and using contrastive pretraining to harness large quantities of unlabelled data.

Similar Items