Instance Segmentation of Indoor Scenes Using a Coverage Loss

By Nathan Silberman, David A. Sontag, Rob Fergus. 2014

99 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Environmental Engineering

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

In Environmental Engineering

Citations in the corpus

Recurrent Instance Segmentation (2015, methodology): “y prediction if and only if P~n ^t=1 t;t^ = 0. The optimal matching, , can be found out efﬁciently by means of the Hungarian algorithm, in a similar vein as in [27]. The coverage loss described in [38] has a similar form, where the predictions were discrete, and the problem was posed as an integer ”
Deep Watershed Transform for Instance Segmentation (2016, methodology): “In addition to AP, we report the mean weighted coverage score introduced in [26] on the validation set.”
Proposal-Free Network for Instance-Level Object Segmentation (2015, neutral): “Recently several approaches which tackle the instance-level object segmentation [14] [3] [30] [39] [34] [13] [6] [5] [27] have emerged.”
Cross Modal Distillation for Supervision Transfer (2015, background): “case of depth images. While the largest labeled RGB dataset ImageNet [6] consists of over a million labeled images, the size of most existing labeled depth datasets is of the order of a few thousands [33, 36, 22]. At the same time there are a large number of unlabeled RGB and depth image pairs. Our ”
Monocular Object Instance Segmentation and Depth Ordering with CNNs (2015, background): “[34] tries to make an optimal cut in the hierarchical segmentation tree to yield object instance regions.”
Pixel-Level Encoding and Depth Layering for Instance-Level Semantic Labeling (2016, background): “Based on proposals that form a segmentation tree, an energy function is constructed in [29] and its solution yields the instance segmentation.”
Instance-Level Segmentation for Autonomous Driving with Deep Densely Connected MRFs (2015, methodology): “[26] makes an optimal cut in a hierarchical segmentation tree to obtain object instance regions.”
Reversible Recursive Instance-Level Object Segmentation (2015, background): “Recently, beyond the traditional object detection [7][24][5][14][4][25] and semantic segmentation tasks [1][18][2][32][16], instance-level object segmentation has attracted much attention [10][11][17][26][31][15].”
Semantic Instance Segmentation with a Discriminative Loss Function (2017, methodology): “Another line of research is to train a network to transform the image into a representation that is clustered into individual instances with a post-processing step [34, 46, 37, 22].”
DeeperLab: Single-Shot Image Parser (2019, methodology): “We note that Covering has been used in several instance segmentation works [92, 74, 76, 93, 78].”
End-to-End Instance Segmentation and Counting with Recurrent Attention (2016, background): “Recent work on instance segmentation [21, 27, 26] formulates complex graphical models, which results in a complex and time-consuming pipeline.”
Semi-convolutional Operators for Instance Segmentation (2018, background): “Among other approaches using a clustering component, [49] leverages a coverage loss and [55,50,51] make use of depth information.”
A Survey of Surface Reconstruction from Point Clouds (2017, methodology): “The approach of [NSF14] is similar to [SDK09], but employs higher-order potentials in a conditional random ﬁeld model to complete extracted contours which enforce scene relationships between wall, ceiling, ﬂoor, and internal plane primitives.”
End-to-End Instance Segmentation with Recurrent Attention (2016, methodology): “[38] is composed of patch-wise features based on deep learning, combined into a segmentation tree.”
Aligning 3D models to RGB-D images of cluttered scenes (2015, background): “There have been corresponding works for RGB-D images studying the problems of object detection [4, 13, 17, 20, 21, 22, 24, 34, 35, 36], semantic segmentation [3, 12, 13, 19, 25, 32, 33], and more recently instance segmentation [13, 33].”
Instance-Level Segmentation with Deep Densely Connected MRFs (2015, methodology): “Our experimental evaluation shows significant improvements over [27] on the challenging KITTI benchmark [8].”
Inferring 3D Object Pose in RGB-D Images (2015, background): “There have been corresponding works for RGB-D images studying the problem of object detection [17, 34, 20, 21, 35, 13, 33, 19, 23, 4], semantic segmentation [3, 12, 18, 24, 31, 13, 32], and more recently instance segmentation [13, 32].”
SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation – A Synthetic Dataset and Baselines (2019, unknown): “Some techniques infer depth-ordering [33, 103, 94, 108], while others are based on a combination of detection and segmentation [104, 37, 38], and yet others combine several over-segmentations [93].”
EfficientPS: Efficient Panoptic Segmentation (2020, background): “C V ] 5 A pr stantial amount of attention in recent recent years (Shotton et al, 2008; Krähenbühl and Koltun, 2011; Silberman et al, 2014; He and Gould, 2014a).”