SegNet swMATH ID: 27575 Software Authors: V. Badrinarayanan, A. Kendall, R. Cipolla Description: SegNet: A deep convolutional encoder-decoder architecture for image segmentation. We present a novel and practical deep fully convolutional neural network architecture for semantic pixel-wise segmentation termed SegNet. This core trainable segmentation engine consists of an encoder network, a corresponding decoder network followed by a pixel-wise classification layer. The architecture of the encoder network is topologically identical to the 13 convolutional layers in the VGG16 network. The role of the decoder network is to map the low resolution encoder feature maps to full input resolution feature maps for pixel-wise classification. The novelty of SegNet lies is in the manner in which the decoder upsamples its lower resolution input feature map(s). Specifically, the decoder uses pooling indices computed in the max-pooling step of the corresponding encoder to perform non-linear upsampling. This eliminates the need for learning to upsample. The upsampled maps are sparse and are then convolved with trainable filters to produce dense feature maps. We compare our proposed architecture with the widely adopted FCN and also with the well known DeepLab-LargeFOV, DeconvNet architectures. This comparison reveals the memory versus accuracy trade-off involved in achieving good segmentation performance. SegNet was primarily motivated by scene understanding applications. Hence, it is designed to be efficient both in terms of memory and computational time during inference. It is also significantly smaller in the number of trainable parameters than other competing architectures. We also performed a controlled benchmark of SegNet and other architectures on both road scenes and SUN RGB-D indoor scene segmentation tasks. We show that SegNet provides good performance with competitive inference time and more efficient inference memory-wise as compared to other architectures. We also provide a Caffe implementation of SegNet and a web demo at this http URL http://mi.eng.cam.ac.uk/projects/segnet/ Homepage: http://mi.eng.cam.ac.uk/projects/segnet/ Source Code: https://github.com/alexgkendall/caffe-segnet Related Software: U-Net; ImageNet; AlexNet; DeepLab; Adam; TensorFlow; DeconvNet; Cityscapes; RefineNet; Matlab; ParseNet; Keras; GitHub; MNIST; PASCAL-Context; OverFeat; MS-COCO; PASCAL VOC; ADE20k; PyTorch Cited in: 24 Documents all top 5 Cited by 72 Authors 3 Fan, Yuwei 3 Ying, Lexing 2 Jia, Fan 2 Lin, Lin 2 Liu, Jun 2 Tai, Xuecheng 2 Zepeda-Núñez, Leonardo 1 Bassir, David Hicham 1 Burgard, Wolfram 1 Cao, Yan 1 Caseria, Brendan J. 1 Chakraborty, Chandan 1 Chen, Guoxiong 1 Chen, Xilin 1 Chung, Tsz Shun Eric 1 Dong, Bin 1 Feliu-Fabà, Jordi 1 Feng, Runhai 1 Gams, Andrej 1 Grana, Dario 1 Guo, Jianyuan 1 Guo, Ling 1 He, Lingxiao 1 Hou, Zhongjun 1 Hu, Ruimeng 1 Huang, Lang 1 Karniadakis, George Em 1 Khatri, Rajendra K. C. 1 Lardy, Jean-Pierre 1 Leung, Wingtat 1 Li, Annan 1 Li, Haiqing 1 Liu, Hui 1 Lou, Yifei 1 Lu, Lu 1 Luo, Jiebo 1 Mauch, Lukas 1 Mercadier, Mathieu 1 Mohan, Rohit 1 Morimoto, Jun 1 Mosegaard, Klaus 1 Mukerji, Tapan 1 Oberai, Assad A. 1 Orozco Bohorquez, Cindy 1 Pahič, Rok 1 Patel, Dhruv V. 1 Pun, Saimang 1 Qi, Mengshi 1 Ridge, Barry 1 Saha, Monjoy 1 Shang, Feifei 1 Sun, Yujuan 1 Sun, Zhenan 1 Ude, Aleš 1 Valada, Abhinav 1 Van Biesbroeck, Antoine 1 Wang, Chunlai 1 Wang, Detao 1 Wang, Jingdong 1 Wang, Yunhong 1 Xiao, Guanghua 1 Yang, Bin 1 Yuan, Yuhui 1 Zabaras, Nicholas J. 1 Zhang, Caiming 1 Zhang, Dongkun 1 Zhang, Haimiao 1 Zhang, Qi 1 Zhang, Xiaofeng 1 Zhang, Zecheng 1 Zhao, Feng 1 Zhu, Yinhao all top 5 Cited in 17 Serials 3 Journal of Computational Physics 3 IEEE Transactions on Image Processing 2 International Journal of Computer Vision 2 Inverse Problems and Imaging 2 Mathematical Geosciences 1 Journal of the Franklin Institute 1 Information Sciences 1 Journal of Computational and Applied Mathematics 1 Neural Networks 1 European Journal of Operational Research 1 Quantitative Finance 1 Multiscale Modeling & Simulation 1 Analysis and Applications (Singapore) 1 International Journal of Computational Methods 1 Journal of the Operations Research Society of China 1 SIAM/ASA Journal on Uncertainty Quantification 1 Research in the Mathematical Sciences all top 5 Cited in 16 Fields 17 Computer science (68-XX) 5 Numerical analysis (65-XX) 4 Partial differential equations (35-XX) 4 Statistics (62-XX) 4 Information and communication theory, circuits (94-XX) 3 Probability theory and stochastic processes (60-XX) 2 Fluid mechanics (76-XX) 2 Geophysics (86-XX) 2 Game theory, economics, finance, and other social and behavioral sciences (91-XX) 2 Biology and other natural sciences (92-XX) 1 Mathematical logic and foundations (03-XX) 1 Dynamical systems and ergodic theory (37-XX) 1 Calculus of variations and optimal control; optimization (49-XX) 1 Mechanics of deformable solids (74-XX) 1 Statistical mechanics, structure of matter (82-XX) 1 Operations research, mathematical programming (90-XX) Citations by Year