Detecting Actions, Poses, and Objects with Relational Phraselets

By Chaitanya Desai, Deva Ramanan. 2012

191 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Cross-View Action Modeling, Learning, and Recognition (2014, background): “[2] learns a deformable part model (DPM) [4] that estimates both human poses and object locations.”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, background): “The context information inves tigated in previous works includes regions surrounding objects [9, 12, 22], object-scene interaction [ 13], and the presence, location, orientation and size relationship among ob jects [3, 57, 58, 11, 41, 22, 49, 13, 61, 12, 59, 40, 10, 45, 51].”
Strong Appearance and Expressive Spatial Models for Human Pose Estimation (2013, methodology): “Part detectors are either trained independently [17, 22] or jointly with the rest of the model [34, 7].”
Differentiable Scene Graphs (2019, methodology): “Visual Relationships: Earlier work aimed to leverage visual relationships for improving detection [33], action recognition and pose estimation [6], semantic image segmentation [9] or detection of human-object interactions”
Detection and Tracking of Occluded People (2014, methodology): “2010), using a large number of part detectors (Poselets) (Bourdev and Malik 2009), detection of interactions between persons and objects (Desai and Ramanan 2012), and careful reasoning about association of image evidence to detection hypotheses (Leibe et al.”
Know More Say Less: Image Captioning Based on Scene Graphs (2019, background): “Earlier approaches explore speciﬁc types of relationships (e.g., spatial relations such as “below”, “above”) and often utilize them to facilitate other tasks [48], [49].”
Learning Human-Object Interactions by Graph Parsing Neural Networks (2018, methodology): “Early work in HOI understanding studied Bayesian model [15, 16], utilized contextual relationship between human and objects [47–49], learned structured representations with spatial interaction and context [8], exploited compositional models [9], or referred to a set of HOI exemplars [19].”
Learning Actionlet Ensemble for 3D Human Action Recognition (2014, methodology): “Poselet has been successfully applied in action recognition by mining discriminative appearance patterns to classify actions [21].”
Learning People Detectors for Tracking in Crowded Scenes (2013, background): “In a similar spirit, approaches to object detection using visual phrases [10] and detection of person-object interactions [7] have demonstrated that detecting a constellation of objects can be easier than detecting each object alone.”
Multi-task Recurrent Neural Network for Immediacy Prediction (2015, methodology): “Both holistic models [28, 11, 19, 20, 33] and local models [31, 8, 39, 6, 9, 30, 24, 10, 26, 14, 29, 35, 25, 1, 21, 4] were used for estimating the pose of a single person.”
Learning Everything about Anything: Webly-Supervised Visual Concept Learning (2014, methodology): “For example, in the case of phraselets [12] and attributes [16, 23], new annotations had to be added to all the PASCAL VOC images.”
Poselet Conditioned Pictorial Structures (2013, methodology): “Our model is related to recent work aiming to increase the flexibility of the PS approach by jointly training a mixture of tree-structured PS models [14, 26, 7].”
Analyzing 3D Objects in Cluttered Images (2012, methodology): “Learning local mixtures: We use the clustering algorithm described in [8, 4] to learn local part mixtures.”
Multi-cue Zero-Shot Learning with Strong Supervision (2016, background): “The notion of semantic parts plays a central role in domains such as human pose estimation [49, 6], action recognition [9, 37, 7] and face detection [55].”
Learning to Detect Human-Object Interactions (2017, methodology): “Results produced in these works were evaluated on either action classification [34, 33, 6, 25, 4, 5, 28, 13], object detection [34], or human pose estimation [34, 5]; none of them were directly evaluated on HOI detection.”
Recurrent Network Models for Human Dynamics (2015, background): “Both occluded and non-occluded body joints are required to be detected correctly: the occluder’s appearance often times contains useful information regarding the location of an occluded body joint [5].”
DeepID-Net: Deformable deep convolutional neural networks for object detection (2014, background): “The context information investigated in literature includes regions surrounding objects [5, 8, 13], object-scene interaction [9, 20], and the presence, location, orientation and size relationship among objects [2, 48, 49, 7, 31, 13, 40, 9, 53, 8, 50, 30, 6, 35, 45].”
Recognizing Actions Through Action-Specific Person Detection (2015, background): “lem of action detection both in images [8], [17], [24] and videos [13], [30], [39].”
Scene Graph Generation from Objects, Phrases and Region Captions (2017, background): “In the early days, most works targeted specific types of phrases [6, 9] or used visual phrases to improve other tasks [36, 16, 25, 35].”