Efficient Human Pose Estimation from Single Depth Images

By Jamie Shotton, Ross B. Girshick, Andrew William Fitzgibbon, Toby Sharp, Mat Cook, Mark Finocchio, Richard Moore, Pushmeet Kohli, Antonio Criminisi, Alex Kipman and 1 more. 2013

462 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

Learning from Simulated and Unsupervised Images through Adversarial Training (2016, methodology): “We use the NYU hand pose dataset [38] that contains 72 , 757 training frames and 8 , 251 testing frames captured by 3 Kinect cameras – one frontal and 2 side views.”
Lifting Object Detection Datasets into 3D (2015, methodology): “Synthetic datasets were also successfully used in Kinect [60], where the inputs can also be rendered realistically (e.g. depth maps).”
Playing for Data: Ground Truth from Computer Games (2016, methodology): “Rendered depth maps of parametric models have been used prominently in training leading systems for human pose estimation and hand tracking [42, 43].”
Learning Depth from Single Monocular Images Using Deep Convolutional Neural Fields (2015, background): “Many challenging computer vision problems have proven to beneﬁt from the incorporation of depth information, to name a few, semantic labellings [1], pose estimations [2].”
Don't just listen, use your imagination: Leveraging visual common sense for non-visual tasks (2015, methodology): “For instance, computer graphics models were used to synthesize data to learn human pose [38] and chair models [2].”
Dense Human Body Correspondences Using Convolutional Networks (2015, background): “Popular skeleton extraction methods from single-view 3D captures such as [38, 46, 43] often have difficulties resolving this ambiguity.”
Anticipating Human Activities Using Object Affordances for Reactive Robotic Response (2013, background): “Recent availability of inexpensive RGB-D sensors has enabled significant improvement in scene modeling [30], [31] and estimation of human poses [32], [33].”
SUN RGB-D: A RGB-D scene understanding benchmark suite (2015, background): “Meanwhile, the recent arrival of affordable depth sensors in consumer markets enables us to acquire reliable depth maps at a very low cost, stimulating breakthroughs in several vision tasks, such as body pose recognition [56, 58], intrinsic image estimation [4], 3D modeling [27] and SfM reconstructi”
Hobbit, a care robot supporting independent living at home: First prototype and lessons learned (2016, background): “[31,32]) have become computationally cheap and readily available at real-time performance on conventional computers.”
Body Parts Dependent Joint Regressors for Human Pose Estimation in Still Images (2014, unknown): “Index Terms—Human pose estimation, fashion, random forest, regression, classification F”
Group-Wise Correlation Stereo Network (2019, background): “Accurate depth sensing plays an important role in many computer vision applications like odometry, robot navigation, pose estimation, and object detection [10, 28, 6, 16].”
Sliding Shapes for 3D Object Detection in Depth Images (2014, methodology): “Meanwhile, the availability of inexpensive RGB-D sensors, such as Microsoft Kinect, Apple PrimeSense, Intel RealSense, and Google Project Tango, has greatly simplified some common challenges in vision and enabled breakthroughs for several tasks, such as body pose estimation [7, 8], intrinsic image [”
Soccer on Your Tabletop (2018, background): “Analysis and reconstruction of people from depth sensors is an active area of research [45, 3], but the use of depth sensors in outdoor scenarios is limited because of the interference with abundant natural light.”
Deep Neural Decision Forests (2015, methodology): “In such a way, random forests could be used as out-of-the-box classifiers for many computer vision tasks such as image classification [3] or semantic segmentation [5, 32], where the input space (and corresponding data representation) they operate on is typically predefined and left unchanged.”
Deep Kinematic Pose Regression (2016, background): “Some approaches represent the object pose in a non-parametric way (as a number of points) and directly learn the pose from data [28,27,5].”