Region-based Segmentation and Object Detection

By Stephen Gould, Tianshi Gao, Daphne Koller. 2009

269 citations recorded by the source corpus.

Object identification and multi-object picture separation are two firmly related processes and it can be enhanced when understood jointly by supporting data from one assignment to the next. Be that as it may, current best in object models are different portrayal for each space creation joint objects and leaving the categorization of numerous part of the scene uncertain. Picture element appearance highlights enable us to do well on classifying formless foundation classes, while the express portrayal of districts encourage the calculation of increasingly complex highlights essential for object detection. Vitally, our model gives a solitary bound together portrayal of the scene we clarify each picture elements of image and authorize it contains in the web between every random variable in our model.

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

You Only Look Once: Unified, Real-Time Object Detection (2015, background): “These classifiers or localizers are run either in sliding window fashion over the whole image or on some subset of regions in the image [34, 15, 38].”
K-Net: Towards Unified Image Segmentation (2021, background): “There are also pioneer attempts [19, 29, 51, 64] to joint the two segmentation tasks for more comprehensive scene understanding.”
Higher Order Conditional Random Fields in Deep Neural Networks (2015, methodology): “[15] used semantic segmentation to propose regions to detect.”
DeeperLab: Single-Shot Image Parser (2019, background): “Since then, there have been several works aiming to jointly perform detection and segmentation for whole scene understanding with AND-OR graphs [9, 10], Exemplars [11, 12, 13], or Conditional Random Fields [14, 15, 16, 17, 18, 19].”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Relating Things and Stuff via Object Property Interactions (methodology): “Our MAP inference algorithm takes only a few seconds per image in average using a single core as opposed to five minutes by Gould et al. [22].”
Efficient Exact Inference for 3D Indoor Scene Understanding (2012, background): “Most outdoor semantic scene understanding approaches produce either only qualitative results [3] or a mild level of understanding in the form of semantic labels [14, 15], object detections [16] or rough 3D [17, 18].”
Relating Things and Stuff via ObjectProperty Interactions (2014, background): “[22] proposed a random field model incorporating both stuff-stuff, thing-stuff, and thing-horizon relationships.”
Resolving Language and Vision Ambiguities Together: Joint Segmentation & Prepositional Attachment Resolution in Captioned Scenes (2016, background): “In Section 4, we describe each of the modules in detail, making concrete their respective features, and other details.”
Object segmentation by alignment of poselet activations to image contours (2011, background): “Similar in spirit, but focusing on a complex graphical model that makes more use of context, is the work in [12].”
Top-down and bottom-up cues for scene text recognition (2012, background): “There have also been significant attempts at addressing all these tasks jointly [14, 16, 20].”
Constrained parametric min-cuts for automatic object segmentation (2010, background): “Bagon et al. [3] also use EM but to estimate a sophisticated self-similarity energy.”
LatentGNN: Learning Efficient Non-local Relations for Visual Recognition (2019, methodology): “Latent Variable Models and Kernel Learning The idea of using latent variables to capture long-range dependency has been widely explored in the literature of probabilistic graphical model (Koller et al., 2009).”
Relating Things and Stuff by High-Order Potential Modeling (2012, methodology): “Our MAP inference algorithm takes only a few seconds per image in average as opposed to five minutes in [8].”
Object detection by labeling superpixels (2015, background): “[22, 58, 59] infer whether an off-the-shelf detection is right or wrong by jointly optimize the detection and segmentation.”
What, Where and How Many? Combining Object Detectors and CRFs (2010, methodology): “Also, [6] focussed only on two classes (cars and pedestrians), while we handle many types of objects ( e.g .”
CPMC: Automatic Object Segmentation Using Constrained Parametric Min-Cuts (2012, background): “Semantic segmentation algorithms [11], [12], [13], [14] incorporate the outputs of these object detectors, and may need to mediate the transition between the rectangular regions produced by the detector and the desired free-form regions that align with object boundaries.”
Analyzing Semantic Segmentation Using Hybrid Human-Machine CRFs (2013, methodology): “Recent works [12, 40, 16, 23], have thus pushed on holistic scene understanding models for among other things, improved semantic segmentation.”
BlitzNet: A Real-Time Deep Network for Scene Understanding (2017, unknown): “The idea of joint semantic segmentation and object detection was investigated ﬁrst for shallow approaches in [3, 18, 29, 7].”
Hough Regions for Joining Instance Localization and Segmentation (2012, background): “[12] jointly estimate detection and segmentation in a unified optimization framework, however with an approximation of the inference step, since their cost formulation is intractable otherwise.”