Contextual Models for Object Detection Using Boosted Random Fields

By Antonio Torralba, Kevin P. Murphy, William T. Freeman. Neural Information Processing Systems, 2004

408 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/a5f39edf5d270c6fd67d8a1ffeab2cc357deb118

How researchers see this paper

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

Citations in the corpus

Latent-Dynamic Discriminative Models for Continuous Gesture Recognition (2007, background): “introduced Boosted Random Fields, a model that combines local and global image information for contextual object recognition [24].”
Building a Large-scale Multimodal Knowledge Base System for Answering Visual Queries (2015, methodology): “While MRFs have been widely used in a variety of vision tasks [9, 26, 27, 46], applying them to a large-scale KB framework means that we need to conquer the challenge of scalable learning and inference.”
A Tree-Based Context Model for Object Recognition (2012, background): “Torralba et al. [27] combine boosting and CRFs to first detect easy objects (e.g., a monitor) and pass the contextual information to detect other more difficult objects (e.g., a keyboard).”
Discriminative learning of Markov random fields for segmentation of 3D scan data (2005, methodology): “[23] propose boosting random fields for image segmentation, combining ideas from boosting and CRFs.”
Detecting Object Boundaries Using Low-, Mid-, and High-level Information (2007, methodology): “Conditional Markov random fields models [10, 16] have been used to enforce local consistency between neighboring structures.”
Using Combination of Statistical Models and Multilevel Structural Information for Detecting Urban Areas From a Single Gray-Level Image (2007, background): “[40], the authors design a unary potential using boosting.”
Beyond Actions: Discriminative Models for Contextual Group Activities (2010, unknown): “Some work clearly demonstrates that contextual information could improve the performance of weak local detectors [19, 39, 30, 5, 45, 51].”
Conditional models for contextual human motion recognition (2006, background): “Therefore, they cannot accommodate multiple overlapping features of the observation or long-range dependencies between state indexes and observation indexes at multiple timesteps because the inference problem in this case becomes intractable.”
Latent hierarchical structural learning for object detection (2010, background): “The nodes can move spatially to allow both local and global shape deformations.”
Learning Aligned Cross-Modal Representations from Weakly Aligned Data (2016, unknown): “techniques [10] have been developed to learn classifiers from a single or a few examples, mostly by reusing classifier parameters [11], using contextual information [27, 18] or sharing part detectors [3].”
G-CNN: An Iterative Grid Based Object Detector (2015, methodology): “Prior to CNN: For many years the problem of object detection was approached by techniques involving sliding window and classification [20, 18].”
Toward Holistic Scene Understanding: Feedback Enabled Cascaded Classification Models (2012, methodology): “Torralba et al. [19] introduced Boosted Random Fields to model object dependency, which used boosting to learn the graph structure and local evidence of a conditional random field.”
Contextual guidance of eye movements and attention in real-world scenes: the role of global features in object search. (2006, unknown): “In Torralba et al. (2003b, 2004) local objects are used to focus computations into image regions likely to contain a target object.”
Geometric context from a single image (2005, background): “[29, 30] has been very Figure 1: Geometric context from a single image: ground (green), sky (blue), vertical regions (red) subdivided into planar orientations (arrows) and non-planar solid (’x’) and porous (’o’).”
Cross-Modal Scene Networks (2016, background): “ques [11] have been developed to learn classifiers from a single or a few examples, mostly by reusing classifier parameters [13], using contextual information [20], [30] or sharing part detectors [4].”
Context models and out-of-context objects (2012, background): “One way to incorporate this idea in object recognition is to learn how likely each object appears in a specific scene category.”
Multi-Class Segmentation with Relative Location Prior (2008, methodology): “[28] use a series of boosting rounds in which easier objects are found first, and local context used to help in the detection of harder objects during later rounds.”
Guiding model search using segmentation (2005, methodology): “[25] use boosted random fields for object detection with contextual information.”