Self-Supervised Learning for Seismic Image Segmentation From Few-Labeled Samples

Autor(es)

Monteiro, Bruno A. A. ; Oliveira, Hugo ; dos Santos, Jefersson A.

• Tipo de documento: Artigo Científico
• Fonte: IEEE Geoscience and Remote Sensing Letters; v. 19, p. 5-pg., 2022-01-01.

Resumo

Current deep learning methods for interpreting seismic images require large amounts of labeled data, and due to strategic and economic interests, these data are not plenty available. In this scenario, seismic interpretation can benefit from self-supervised learning (SSL) by relying on prior training without manually annotated labels within the target data domain and subsequent fine-tuning with few shots. To demonstrate the potential of such an approach, we conducted experiments with three classic context-based pretext tasks: rotation, jigsaw, and frame order prediction. Our results for 1, 5, 10, and 20 shots showed significant improvement for mean Intersection-over-Union (mIoU) measurements for semantic segmentation in most scenarios, outperforming the baseline method in 38% in the one-shot scenario for the F3 Netherlands Dataset and 16.4% in the New Zealand Parihaka dataset, and this gap grows even higher after performing ensemble modeling. These experiments suggest that applying SSL methods can also bring great benefits in seismic interpretation when few labeled data are available. (AU)

• Processo FAPESP: 20/06744-5 - Deep learning e representações intermediárias para análise de imagens pediátricas
• Beneficiário: Hugo Neves de Oliveira
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado
• Processo FAPESP: 15/22308-2 - Representações intermediárias em Ciência Computacional para descoberta de conhecimento
• Beneficiário: Roberto Marcondes Cesar Junior
• Modalidade de apoio: Auxílio à Pesquisa - Temático