Object Segmentation
Deep Joint Task Learning for Generic Object Extraction(NIPS2014)
- homepage: http://vision.sysu.edu.cn/projects/deep-joint-task-learning/
- paper: http://ss.sysu.edu.cn/~ll/files/NIPS2014_JointTask.pdf
- code: https://github.com/xiaolonw/nips14_loc_seg_testonly
- dataset: http://objectextraction.github.io/
Fully Convolutional Networks for Semantic Segmentation
- keywords: deconvolutional layer, crop layer
- arxiv: http://arxiv.org/abs/1411.4038
- slides: https://docs.google.com/presentation/d/1VeWFMpZ8XN7OC3URZP4WdXvOGYckoFWGVN7hApoXVnc
- slides: http://tutorial.caffe.berkeleyvision.org/caffe-cvpr15-pixels.pdf
- github: https://github.com/BVLC/caffe/wiki/Model-Zoo#fcn
- notes: http://zhangliliang.com/2014/11/28/paper-note-fcn-segment/
Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs(“DeepLab”)
- intro: “adopted a more simplistic approach for maintaining resolution by removing the stride in the layers of FullConvNet, wherever possible. Following this, the FullConvNet predicted output is modeled as a unary term for Conditional Random Field (CRF) constructed over the image grid at its original resolution. With labelling smoothness constraint enforced through pair-wise terms, the per-pixel classification task is modeled as a CRF inference problem.”
- arXiv: http://arxiv.org/abs/1412.7062
- github: https://bitbucket.org/deeplab/deeplab-public/
Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation(“DeepLab”)
- arXiv: http://arxiv.org/abs/1502.02734
- bitbucket: https://bitbucket.org/deeplab/deeplab-public/
Hypercolumns for object segmentation and fine-grained localization (CVPR 2015)
Conditional Random Fields as Recurrent Neural Networks(ICCV2015. Oxford/Stanford/Baidu)
- intro: “proposed a better approach where the CRF constructed on image is modeled as a Recurrent Neural Network (RNN). By modeling the CRF as an RNN, it can be integrated as a part of any Deep Convolutinal Net making the system efficient at both semantic feature extraction and fine-grained structure prediction. This enables the end-to-end training of the entire FullConvNet + RNN system using the stochastic gradient descent (SGD) algorithm to obtain fine pixel-level segmentation.”
- arXiv: http://arxiv.org/abs/1502.03240
- homepage: http://www.robots.ox.ac.uk/~szheng/CRFasRNN.html
- github: https://github.com/torrvision/crfasrnn
- demo: http://www.robots.ox.ac.uk/~szheng/crfasrnndemo
Learning to Segment Object Candidates
Proposal-free Network for Instance-level Object Segmentation
Semantic Image Segmentation via Deep Parsing Network
- paper: http://arxiv.org/abs/1509.02634
- homepage: http://personal.ie.cuhk.edu.hk/~lz013/projects/DPN.html
Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation(NIPS 2015)
- paper: http://arxiv.org/abs/1506.04924
- project[paper+code]: http://cvlab.postech.ac.kr/research/decouplednet/
- github: https://github.com/HyeonwooNoh/DecoupledNet
Learning Deconvolution Network for Semantic Segmentation
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Robust Semantic Pixel-Wise Labelling
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
- homepage: http://mi.eng.cam.ac.uk/projects/segnet/
- arXiv: http://arxiv.org/abs/1511.00561
- github: https://github.com/alexgkendall/caffe-segnet
- tutorial: http://mi.eng.cam.ac.uk/projects/segnet/tutorial.html
SegNet: Pixel-Wise Semantic Labelling Using a Deep Networks
Recurrent Instance Segmentation
- arXiv: http://arxiv.org/abs/1511.08250
- homepage: http://romera-paredes.com/recurrent-instance-segmentation
Instance-aware Semantic Segmentation via Multi-task Network Cascades
- intro: “1st-place winner of MS COCO 2015 segmentation competition”
- arxiv: http://arxiv.org/abs/1512.04412
Semantic Object Parsing with Graph LSTM
