TY - GEN
T1 - An adaptive sampling scheme to efficiently train fully convolutional networks for semantic segmentation
AU - Berger, Lorenz
AU - Eoin, Hyde
AU - Cardoso, M. Jorge
AU - Ourselin, Sébastien
PY - 2018/8/21
Y1 - 2018/8/21
N2 - Deep convolutional neural networks (CNNs) have shown excellent performance in object recognition tasks and dense classification problems such as semantic segmentation. However, training deep neural networks on large and sparse datasets is still challenging and can require large amounts of computation and memory. In this work, we address the task of performing semantic segmentation on large data sets, such as three-dimensional medical images. We propose an adaptive sampling scheme that uses a-posterior error maps, generated throughout training, to focus sampling on difficult regions, resulting in improved learning. Our contribution is threefold: (1) We give a detailed description of the proposed sampling algorithm to speed up and improve learning performance on large images. (2) We propose a deep dual path CNN that captures information at fine and coarse scales, resulting in a network with a large field of view and high resolution outputs. (3) We show that our method is able to attain new state-of-the-art results on the VISCERAL Anatomy benchmark.
AB - Deep convolutional neural networks (CNNs) have shown excellent performance in object recognition tasks and dense classification problems such as semantic segmentation. However, training deep neural networks on large and sparse datasets is still challenging and can require large amounts of computation and memory. In this work, we address the task of performing semantic segmentation on large data sets, such as three-dimensional medical images. We propose an adaptive sampling scheme that uses a-posterior error maps, generated throughout training, to focus sampling on difficult regions, resulting in improved learning. Our contribution is threefold: (1) We give a detailed description of the proposed sampling algorithm to speed up and improve learning performance on large images. (2) We propose a deep dual path CNN that captures information at fine and coarse scales, resulting in a network with a large field of view and high resolution outputs. (3) We show that our method is able to attain new state-of-the-art results on the VISCERAL Anatomy benchmark.
UR - http://www.scopus.com/inward/record.url?scp=85052868214&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-95921-4_26
DO - 10.1007/978-3-319-95921-4_26
M3 - Conference contribution
AN - SCOPUS:85052868214
SN - 9783319959207
T3 - Communications in Computer and Information Science
SP - 277
EP - 286
BT - Medical Image Understanding and Analysis - 22nd Conference, Proceedings
PB - Springer Verlag
T2 - 22nd Conference on Medical Image Understanding and Analysis, MIUA 2018
Y2 - 9 July 2018 through 11 July 2018
ER -