What, Where and How Many? Combining Object Detectors and CRFs

By Lubor Ladicky, Paul Sturgess, Alahari Karteek, Chris Russell, Philip H. S. Torr. 2010

369 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

Pixelwise Instance Segmentation with a Dynamically Instantiated Network (2017, methodology): “This detection potential is similar to the one previously proposed by [25], [41], [45] and [48], but formulated for the differentiable mean field inference algorithm.”
Simultaneous Detection and Segmentation (2014, methodology): “[22] use object detections as higher order potentials in a CRF-based segmentation system: all pixels in the foreground of a detected object are encouraged to share the category label of the detection.”
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation (2015, methodology): “, Random Forest [26], [27] or Boosting [28], [29] to predict the class probabilities of the center pixel.”
Higher Order Conditional Random Fields in Deep Neural Networks (2015, background): “Conditional Random Fields (CRFs) have been the most common framework, and various rich and expressive models [27, 29, 48], based on higher order clique potentials, have been developed to improve segmentation performance.”
Recurrent Pixel Embedding for Instance Grouping (2017, neutral): “Common approaches to instance segmentation ﬁrst generate region proposals or class-agnostic bounding boxes, segment the foreground objects within each proposal and classify the objects in the bounding box [92, 53, 31, 12, 17, 56, 34].”
Bottom-up Instance Segmentation using Deep Higher-Order CRFs (2016, methodology): “Subsequent improvements included introducing densely-connected pairwise potentials [18], using higher order potentials encouraging consistency over cliques larger than two pixels [17, 35], modelling the co-occurrence of labels [20, 35] and utilising the results of object detectors [21, 33, 36].”
segDeepM: Exploiting segmentation and context in deep neural networks for object detection (2015, methodology): “The standard approach has been to use detection features as unary potentials in an MRF [14], or as candidate bounding boxes for holistic MRFs [26, 16].”
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].”
Machine Learning and Radiology (2012, background): “In recent years, joint categorization and segmentation (JCaS) has become a hot topic in computer vision (Ladicky et al., 2010; Singaraju and Vidal, 2011).”
Bottom-Up Segmentation for Top-Down Detection (2013, background): “This is typically done in the form of unary features for an MRF [19], or as candidate bounding boxes for holistic MRFs [33] [21].”
Understanding web images by object relation network (2012, neutral): “The probabilistic graphical model can be a hierarchical structure [27, 14], a directed graphical model [26], or a conditional random field [10, 20, 11, 12].”
Instance-Level Segmentation with Deep Densely Connected MRFs (2015, methodology): “[15, 26] proposed an MRF model to identify which detections are true positive, and output pixelwise class labels, thus performing instance segmentation.”
Semi-convolutional Operators for Instance Segmentation (2018, background): “In general, they implement a multi-stage pipeline that first generates region proposals or class agnostic boxes, and then classifies them [29,20,7,42,10,43,32].”
Superparsing (2010, background): “Many approaches to this problem have been proposed recently, including ones that estimate labels pixel by pixel (He et al. 2004; Ladicky et al. 2010; Shotton et al. 2006, 2008), ones that aggregate features over segmentation regions (Galleguillos et al.”