Pose Machines: Articulated Pose Estimation via Inference Machines

By Varun Ramakrishna, Daniel Munoz, Martial Hebert, J. Bagnell, Yaser Sheikh. 2014

263 citations recorded by the source corpus.

State-of-the-art approaches for articulated human pose estimation are rooted in parts-based graphical models. These models are often restricted to tree-structured representations and simple parametric potentials in order to enable tractable inference. However, these simple dependencies fail to capture all the interactions between body parts. While models with more complex interactions can be defined, learning the parameters of these models remains challenging with intractable or approximate inference. In this paper, instead of performing inference on a learned graphical model, we build upon the inference machine framework and present a method for articulated human pose estimation. Our approach incorporates rich spatial interactions among multiple parts and information across parts of different scales. Additionally, the modular framework of our approach enables both ease of implementation without specialized optimization solvers, and efficient inference. We analyze our approach on two challenging datasets with large pose variation and outperform the state-of-the-art on these benchmarks.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Human-Computer Interaction

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

In Human-Computer Interaction

Citations in the corpus

Realtime Multi-person 2D Pose Estimation Using Part Affinity Fields (2016, methodology): “The convolutional pose machines architecture proposed by Wei et al. [38] used a multi-stage architecture based on a sequential prediction framework [29]; iteratively incorporating global context to reﬁne part conﬁ-dence maps and preserving multimodal uncertainty from previous iterations.”
Convolutional Pose Machines (2016, methodology): “The pose machine [29] is shown in insets (a) and (b), and the corresponding convolutional networks are shown in insets (c) and (d).”
Stacked Hourglass Networks for Human Pose Estimation (2016, background): “[18] build on the work of multi-stage pose machines [26] but now with the use of ConvNets for feature extraction.”
Human Pose Estimation with Iterative Error Feedback (2015, methodology): “Stacked inference methods [31, 44, 45, 41] are another related family of methods.”
DeeperCut: A Deeper, Stronger, and Faster Multi-person Pose Estimation Model (2016, background): “Recent work aimed to incorporate convolutional detectors into part-based models [9] or design stronger detectors by combining the detector output with location-based features [20].”
CliqueCNN: Deep Unsupervised Exemplar Learning (2016, methodology): “For completeness, we compare with a fully supervised state-of-the-art approach for pose estimation [18].”
Fashion Landmark Detection in the Wild (2016, methodology): “It shows comparable results to [11] and outperforms several recent works [32,39,30,10], showing that DFA is a general approach for structural prediction problem besides fashion landmark detection.”
Self Adversarial Training for Human Pose Estimation (2017, background): “Early work often tackles the problem using graphical models [2, 13, 22] and random field inference [23, 35] with handcrafted image features.”
An Invitation to Imitation (2015, background): “These include examples from predicting semantic labels for images [Munoz et al., 2010 ], identifying human poses in images and video [Ramakrishna et al., 2014 ], summarizing documents with the SCP algorithm [Ross et al., 2013 c], and a broad range of Natural Language Processing Tasks [Daumé III et…”
Uncertainty-Driven 6D Pose Estimation of Objects and Scenes from a Single RGB Image (2016, methodology): “Auto-context has been widely used for segmentation [28, 24, 20, 12], detection [11], and human pose estimation [21, 3], etc.”
Learning to Filter with Predictive State Inference Machines (2015, background): “Instead of simply predicting next observation, we focus on predictive state—a distribution of future observations, as an alternative representation of the beliefs of latent states.”
Dense Semantic Correspondence Where Every Pixel is a Classifier (2015, methodology): “We pose the correspondence problem as a graphical model, where the unary potentials are computed via convolution with the set of exemplar classifiers, and the joint potentials enforce smoothly varying correspondence assignment.”
Recognizing Human Actions as the Evolution of Pose Estimation Maps (2018, methodology): “In [34], a pose machine is proposed to sequentially predict pose estimation maps for body joints, where previous predicted pose estimation maps iteratively improve the estimates in following stages.”
Deep Multitask Architecture for Integrated 2D and 3D Human Sensing (2017, methodology): “Note that the iterative processing frameworks of [33, 15] are not deep, built on parts-based graphical models and random forests, respectively.”
Monocular human pose estimation: A survey of deep learning-based methods (2020, methodology): “…with action recognition on video sequences (Luvizon et al., 2018) Detection-based (1) Patch-based : (Jain et al., 2013; Chen and Yuille, 2014; Ramakrishna et al., 2014) (2) Network design : (Tompson et al., 2015; Bulat and Tzimiropoulos, 2016; Xiao et al., 2018), multi-scale inputs (Raﬁ et…”
Deep Learning-based Human Pose Estimation: A Survey (2020, background): “Therefore, there is a recent growing interest in leveraging heatmaps to represent the joint locations and developing effective CNN architectures for HPE, e.g., References [6, 7, 11, 46, 59, 124, 129, 145, 171, 200, 231, 232, 254, 259, 269, 285].”