Progressive search space reduction for human pose estimation

By Vittorio Ferrari, Manuel Jesús Marín-Jiménez, Andrew Zisserman. 2008

736 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 3D Vision & Pose

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

1. Bernt Schiele
2. Trevor Darrell
3. Luc Van Gool
4. Li Fei-Fei
5. Xiaogang Wang
6. Jitendra Malik
7. Abhinav Kumar Gupta
8. Mohit Bansal

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Xiaogang Wang
3. Jitendra Malik
4. Andrew Zisserman
5. Dacheng Tao
6. Bernt Schiele
7. Ming-Hsuan Yang
8. Trevor Darrell

In 3D Vision & Pose

1. Jitendra Malik
2. Michael J. Black
3. Juergen Gall
4. Bernt Schiele
5. Peter Gehler
6. Mykhaylo Andriluka
7. Cordelia Schmid
8. Alan L. Yuille

Citations in the corpus

• Extreme Clicking for Efficient Object Annotation (2017, background): “For the object, one possible approach is to clamp a small area in the center of the box [23].”
• DeepPose: Human Pose Estimation via Deep Neural Networks (2013, methodology): “Further, the upper-body model trained on FLIC was applied on the whole Buffy dataset [7].”
• Stacked Hourglass Networks for Human Pose Estimation (2016, background): “Early work tackles such difficulties using robust image features and sophisticated structured prediction [1,2,3,4,5,6,7,8,9]: the former is used to produce local interpretations, whereas the latter is used to infer a globally consistent pose.”
• Detect What You Can: Detecting and Representing Objects Using Holistic Models and Body Parts (2014, methodology): “We adopt the widely used measure of PCP (Percentage of Correctly estimated body Parts) [10] to evaluate the object parts localized by our method.”
• Articulated Pose Estimation by a Graphical Model with Image Dependent Pairwise Relations (2014, methodology): “We also do cross-dataset evaluation on Buffy dataset [7] (without training on this dataset) and obtain strong results which shows the ability of our model to generalize.”
• Articulated Human Detection with Flexible Mixtures of Parts (2013, methodology): “We suspect that previous authors either report a single candidate pose per image, or multiple poses that are matched using the code of [7].”
• Recurrent Human Pose Estimation (2016, methodology): “many years approaches have used variations on the pictorial structure model [14], [16] of a combination of part detectors and configuration constraints [2], [15], [30], [34], [44].”
• Strong Appearance and Expressive Spatial Models for Human Pose Estimation (2013, methodology): “As a performance measure we use the common PCP loss [12].”
• Cascaded hand pose regression (2015, methodology): “Hierarchical pose estimation The progressive human pose estimation approach [11] shares a similar idea with our hierarchical regression at a high abstract level, but differs a lot in details.”
• Using k-Poselets for Detecting People and Localizing Their Keypoints (2014, positive): “We compute the mean AP for all keypoints (column mAP) and also for the arm and leg keypoints for a comparison with Y&R (column mAP ∗ ) The task of keypoint prediction on VAL2 is challenging, especially compared to other pose estimation datasets, such as Buffy Stickmen [12] or the PARSE dataset [21],”
• Figure-ground segmentation improves handled object recognition in egocentric video (2010, methodology): “[8] uses two rectangles as the prior to highlight foreground regions of people in movies.”
• Beyond Physical Connections: Tree Models in Human Pose Estimation (2013, methodology): “The evaluation criterion is the same with [16] for performance comparison.”
• Action Recognition with Exemplar Based 2.5D Graph Matching (2012, methodology): “5D graphs are constructed based on imperfect key-point locations (using the criteria in [39], our key-point detection accuracy is 65.”
• 3D Human pose estimation: A review of the literature and analysis of covariates (2016, background): “A common 2D pose estimation error in the literature is the Percentage of Correctly estimated Parts (PCP) error which measures the percentage of correctly localized body parts [153].”
• Learning hierarchical poselets for human parsing (2011, background): “Buffy dataset [8], PASCAL stickmen dataset [5].”
• "Here's looking at you, kid". Detecting people looking at each other in videos (2011, background): “Determining whether characters are interacting using their eyelines is another step towards a fuller video understanding, and complements recent work on automatic character identification [7, 11, 27], human pose estimation [1, 6, 9, 13, 14, 24], human action recognition [16, 17, 18], and specific in”
• Structured Prediction Cascades (2010, methodology): “We compared the single-frame cascade and the ensemble cascade to a state-of-the-art single-frame pose detector (Ferrari et al. (Ferrari et al., 2008)) and to one of the individual sub-models, modeling torso consistency only (“Torso Only”).”
• Fine-Grained Activity Recognition with Holistic and Pose Based Features (2014, background): “Another line of research explores ways of higher level video encoding in terms of body pose and motion [11,25,21,12].”
• Robust Optimization for Deep Regression (2015, methodology): “In addition, we employ the PCP (percentage of correctly estimated parts) performance measure, which is the standard metric used in human pose estimation [11].”