Application of Deep Neural Networks for Spatial Relation Learning in Seismic Object Detection
عنوان دوره: ششمین همایش ژئوفیزیک اکتشافی نفت
کد مقاله : 2137-NIGS
نویسندگان
1R&D Specialist
2Research and Development Manager
3Reservoir Characterization Lead
چکیده
Deep Learning (DL) is the state-of-the-art Machine Learning (ML) technique which is widely deployed in academia and industry. DL allows automated identification of complicated patterns in large data sets (“big data”). Since the seismic data can be treated as image, there have been many successful applications of DL in this field. In this study, we focus on DL in seismic interpretation, specifically seismic object detection. We present the application of the CNN technique to classify gas chimneys from shallow sediments containing gas. These sediments show similar characteristics as gas chimneys in selected seismic attributes used in conventional ML classification methods. Our work showcases deep learning's exceptional capacity for spatial relationship modeling, enabling accurate seismic object detection even in complex cases where conventional machine learning approaches struggle.
کلیدواژه ها
Title
Application of Deep Neural Networks for Spatial Relation Learning in Seismic Object Detection
Authors
Mohammad Ghasem Fakhari, Ehsan Salehi, Mahdi Saadat
Abstract
Deep Learning (DL) is the state-of-the-art Machine Learning (ML) technique which is widely deployed in academia and industry. DL allows automated identification of complicated patterns in large data sets (“big data”). Since the seismic data can be treated as image, there have been many successful applications of DL in this field. In this study, we focus on DL in seismic interpretation, specifically seismic object detection. We present the application of the CNN technique to classify gas chimneys from shallow sediments containing gas. These sediments show similar characteristics as gas chimneys in selected seismic attributes used in conventional ML classification methods. Our work showcases deep learning's exceptional capacity for spatial relationship modeling, enabling accurate seismic object detection even in complex cases where conventional machine learning approaches struggle.
Keywords
Deep Learning (DL), Convolutional Neural Network (CNN), Gas chimney