DeepLab

Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Dados Bibliográficos

ID23314935
AutoresLiang-Chieh Chen (0000-0001-8564-374X, Google (United States)), George Papandreou (0000-0002-1468-2976, Google (United States)), Iasonas Kokkinos (0000-0002-2606-6476, University College London), Kevin Murphy (0000-0001-8982-3641, Google (United States)), Alan Yuille (0000-0001-5207-9249, Johns Hopkins University), Alan L Yuille
Ano2018
Volume40
Fascículo4
Páginas834-848
Data de publicação2018-04-01
Peer ReviewedSim
Open AccessSim
TipoARTICLE
PeriódicoIEEE Transactions on Pattern Analysis and Machine Intelligence (JOURNAL)
Identificadores do periódicoISSN: 0162-8828 • E-ISSN: 1939-3539
EditoraInstitute of Electrical and Electronics Engineers (IEEE) (PUBLISHER)
DOI10.1109/tpami.2017.2699184
PMID28463186
OpenAlexW2412782625
IdiomaEN
Citações recebidas89
Referências citadas64

In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit. First, we highlight convolution with upsampled filters, or 'atrous convolution', as a powerful tool in dense prediction tasks. Atrous convolution allows us to explicitly control the resolution at which feature responses are computed within Deep Convolutional Neural Networks. It also allows us to effectively enlarge the field of view of filters to incorporate larger context without increasing the number of parameters or the amount of computation. Second, we propose atrous spatial pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales. Third, we improve the localization of object boundaries by combining methods from DCNNs and probabilistic graphical models. The commonly deployed combination of max-pooling and downsampling in DCNNs achieves invariance but has a toll on localization accuracy. We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance. Our proposed "DeepLab" system sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 79.7 percent mIOU in the test set, and advances the results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and Cityscapes. All of our code is made publicly available online.

Artificial neural network · Conditional random field · Contextual image classification · Convolution (computer science) · Convolutional neural network · CRFS · Feature (linguistics) · Feature extraction · Image (mathematics) · Image segmentation · Markov random field · Object detection · Pascal (unit) · Pattern recognition (psychology) · Pooling · Segmentation · Upsampling · Advanced Image and Video Retrieval Techniques · Advanced Neural Network Applications · Artificial Intelligence · Computer Science · Robotics and Sensor-Based Localization