Weakly supervised semantic segmentation with a multi-image model

By Alexander Vezhnevets, Vittorio Ferrari, Joachim M. Buhmann. 2011

179 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

COCO-Stuff: Thing and Stuff Classes in Context (2016, methodology): “We then define the objectness score for a pixel to be the average score over all boxes that contain it, as done in [6, 55].”
DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (2016, neutral): “[69], [70], [71], achieving significantly better results than weakly-supervised pre-DCNN systems such as [72].”
Weakly-and Semi-Supervised Learning of a Deep Convolutional Network for Semantic Image Segmentation (2015, methodology): “Related Work Training semantic image segmentation models using solely image-level labels for is a challenging problem that has attracted much interest in the literature (Duygulu et al., 2002; Verbeek & Triggs, 2007; Vezhnevets & Buhmann, 2010; Vezhnevets et al., 2011; 2012; Xu et al., 2014).”
From image-level to pixel-level labeling with Convolutional Networks (2014, methodology): “The unary potentials of the MIM are initialized with the output of [25].”
What's the Point: Semantic Segmentation with Point Supervision (2015, methodology): “The most common setting is where only image-level labels for the presence or absence of classes are provided during training [37, 38, 35, 26, 28, 40, 29] (Fig.”
Seed, Expand and Constrain: Three Principles for Weakly-Supervised Image Segmentation (2016, background): “Variants of multiple instance learning [1] train with a per-image loss function, while internally maintaining a spatial representation of the image that can be used to produce segmentation masks [37,38,39].”
Weakly-Supervised Dual Clustering for Image Semantic Segmentation (2013, methodology): “Vezhnevets et.al [26] proposed a graphical MIM model and introduced an objectness to distinguish objects from background classes.”
Learning to segment under various forms of weak supervision (2015, methodology): “Dataset: To illustrate the performance of our model, we conduct a rigorous evaluation on the Siftflow data set [21], which has been widely studied [21, 10, 8, 33, 35, 36, 44, 38, 39, 43].”
Weakly Supervised Semantic Segmentation with Convolutional Networks (2014, methodology): “To model these relationships, [24] introduced a graphical model – named Multi-Image Model (MIM) – to connect superpixels from all training images, based on their appearance similarity.”
Weakly Supervised Object Boundaries (2015, background): “Multiple works have addressed weakly supervised learning for object localization [23, 7], object detection [26, 34], or semantic labelling [33, 36, 24].”
Semantic Segmentation based on Stacked Discriminative Autoencoders and Context-Constrained Weakly Supervised Learning (2015, background): “Recently, many efforts have been contributed to this study [1-6].”
Co-Saliency Detection via a Self-Paced Multiple-Instance Learning Framework (2017, background): “Another topic related to co-saliency detection is weakly supervised semantic segmentation (WSSS) [24], [25], [26], [27], [28], [29], whose goal is to assign semantic labels to each image pixel under the weak supervision of image-level annotations .”
Object Region Mining with Adversarial Erasing: A Simple Classification to Semantic Segmentation Approach (2017, background): “Some works [16–19, 27, 31] propose to train the segmentation models by only using image-level labels, which is the simplest supervision for training semantic segmentation models.”
Similarity Constrained Latent Support Vector Machine: An Application to Weakly Supervised Action Classification (2012, background): “Alexe et al. and Vezhnevets et al. [6, 7] have done impressive work on automatic object localization and segmentation from weakly supervised data.”
Exploiting Saliency for Object Segmentation from Image Level Labels (2017, methodology): “The seminal work [45] proposed to use “objectness” map from bounding boxes to guide the semantic segmentation.”
Clothing Co-parsing by Joint Image Segmentation and Labeling (2014, methodology): “We construct a multi-image graphical model by taking the regions as vertices of graph, inspired by [27].”