Monocular 3D Human Pose Estimation Research Articles

Disentangled Diffusion-Based 3D Human Pose Estimation with Hierarchical Spatial and Temporal Denoiser

Recently, diffusion-based methods for monocular 3D human pose estimation have achieved state-of-the-art (SOTA) performance by directly regressing the 3D joint coordinates from the 2D pose sequence. Although some methods decompose the task into bone length and bone direction prediction based on the human anatomical skeleton to explicitly incorporate more human body prior constraints, the performance of these methods is significantly lower than that of the SOTA diffusion-based methods. This can be attributed to the tree structure of the human skeleton. Direct application of the disentangled method could amplify the accumulation of hierarchical errors, propagating through each hierarchy. Meanwhile, the hierarchical information has not been fully explored by the previous methods. To address these problems, a Disentangled Diffusion-based 3D human Pose Estimation method with Hierarchical Spatial and Temporal Denoiser is proposed, termed DDHPose. In our approach: (1) We disentangle the 3d pose and diffuse the bone length and bone direction during the forward process of the diffusion model to effectively model the human pose prior. A disentanglement loss is proposed to supervise diffusion model learning. (2) For the reverse process, we propose Hierarchical Spatial and Temporal Denoiser (HSTDenoiser) to improve the hierarchical modelling of each joint. Our HSTDenoiser comprises two components: the Hierarchical-Related Spatial Transformer (HRST) and the Hierarchical-Related Temporal Transformer (HRTT). HRST exploits joint spatial information and the influence of the parent joint on each joint for spatial modeling, while HRTT utilizes information from both the joint and its hierarchical adjacent joints to explore the hierarchical temporal correlations among joints. Extensive experiments on the Human3.6M and MPI-INF-3DHP datasets show that our method outperforms the SOTA disentangled-based, non-disentangled based, and probabilistic approaches by 10.0%, 2.0%, and 1.3%, respectively.

DGFormer: Dynamic graph transformer for 3D human pose estimation

Despite the significant progress for monocular 3D human pose estimation, it still faces challenges due to self-occlusions and depth ambiguities. To tackle those issues, we propose a novel Dynamic Graph Transformer (DGFormer) to exploit local and global relationships between skeleton joints for pose estimation. Specifically, the proposed DGFormer mainly consists of three core modules: Transformer Encoder (TE), immobile Graph Convolutional Network (GCN), and dynamic GCN. TE module leverages the self-attention mechanism to learn the complex global relationships among skeleton joints. The immobile GCN is responsible for capturing the local physical connections between human joints, while the dynamic GCN concentrates on learning the sparse dynamic K-nearest neighbor interactions according to different action poses. By building the adequately global long-range, local physical, and sparse dynamic dependencies of human joints, experiments on Human3.6M and MPI-INF-3DHP datasets demonstrate that our method can predict 3D pose with lower errors outperforming the recent state-of-the-art image-based performance. Furthermore, experiments on in-the-wild videos demonstrate the impressive generalization abilities of our method. Code will be available at: https://github.com/czmmmm/DGFormer.

TSwinPose: Enhanced monocular 3D human pose estimation with JointFlow

Monocular estimation of 3D human poses is challenging due to ambiguity in depths and partial occlusion. Most recent works define this as a 2D-to-3D lifting task, taking 2D key point sequences and using spatial and temporal relationships. However, prior works focus on capturing spatio-temporal correlations but ignore the motion of joints that is needed for continuous estimation. To extend the potential of 2D-to-3D pose estimation, we propose TSwinPose, which learns multi-scale spatio-temporal representations from 2D key point locations and patterns of motion. The input 2D key point sequences are enhanced by JointFlow, which encodes the motion of each human joint. Based on Swin-Transformer, we designed a temporal domain Swin-Unet structure to model multi-scale spatio-temporal relationships of human joints across different temporal windows. The final 3D pose generated by multi-stage representations is consistent temporally and has a higher accuracy. Experiments conducted on three benchmark datasets, Human3.6M, MPI-INF-3DHP, and HumanEva-I, demonstrate that TSwinPose achieves performance that is on par with state-of-the-art methods. Moreover, the introduction of JointFlow as a plug-in extension enhances performance significantly, particularly benefiting long-term 2D-to-3D lifting human pose estimation methods.

A Review of Human Pose Estimation Methods Based on Monocular Vision

This article is a review of the technology methods and industry applications of monocular human pose estimation, which focuses on the recent development of monocular human pose estimation. With the rapid development of computer vision and machine learning, monocular human pose estimation has become a research direction that has attracted much attention. In this article, we first introduce the relevant concepts and significance of monocular human pose estimation, explain the importance of research in this field. Then, we provide a detailed overview of the technical methods of monocular human pose estimation, including two different approaches: monocular 2D human pose estimation and monocular 3D human pose estimation. For each research method, we discuss its principles, development, and pros and cons. Next, we explore the applications of monocular human pose estimation in various fields such as human-computer interaction, autonomous driving, healthcare, and provide examples to highlight the significance of human pose estimation. Finally, we analyze the current research hotspots and challenges and also look forward to the future directions of monocular human pose estimation. This article aims to provide researchers and practitioners with a comprehensive overview and promote the application and further research of monocular human pose estimation technology.

Extracting spatial knowledge from track and field broadcasts for monocular 3D human pose estimation

Collecting large datasets for investigations into human locomotion is an expensive and labor-intensive process. Methods for 3D human pose estimation in the wild are becoming increasingly accurate and could soon be sufficient to assist with the collection of datasets for analysis into running kinematics from TV broadcast data. In the domain of biomechanical research, small differences in 3D angles play an important role. More precisely, the error margins of the data collection process need to be smaller than the expected variation between athletes. In this work, we propose a method to infer the global geometry of track and field stadium recordings using lane demarcations. By projecting estimated 3D skeletons back into the image using this global geometry, we show that current state-of-the-art 3D human pose estimation methods are not (yet) accurate enough to be used in kinematics research.

EMHIFormer: An Enhanced Multi-Hypothesis Interaction Transformer for 3D human pose estimation in video

Monocular 3D human pose estimation is a challenging task because of depth ambiguity and occlusion. Recent methods exploit spatio-temporal information and generate different hypotheses for simulating diverse solutions to alleviate these problems. However, these methods do not fully extract spatial and temporal information and the relationship of each hypothesis. To ease these limitations, we propose EMHIFormer (Enhanced Multi-Hypothesis Interaction Transformer) to model 3D human pose with better performance. In detail, we build connections between different Transformer layers so that our model is able to integrate spatio-temporal information from the previous layer and establish more comprehensive hypotheses. Furthermore, a cross-hypothesis model consisting of a parallel Transformer is proposed to strengthen the relationship between various hypotheses. We also design an enhanced regression head which adaptively adjusts the channel weights to export the final 3D human pose. Extensive experiments are conducted on two challenging datasets: Human3.6M and MPI-INF-3DHP to evaluate our EMHIFormer. The results show that EMHIFormer achieves competitive performance on Human3.6M and state-of-the-art performance on MPI-INF-3DHP. Compared with the closest counterpart, MHFormer, our model outperforms it by 0.6% P-MPJPE and 0.5% MPJPE on Human3.6M dataset and 46.0% MPJPE on MPI-INF-3DHP.

A monocular 3D human pose estimation approach for virtual character skeleton retargeting

A monocular 3D human pose estimation approach for virtual character skeleton retargeting

Dual Networks Based 3D Multi-Person Pose Estimation From Monocular Video.

Monocular 3D human pose estimation has made progress in recent years. Most of the methods focus on single persons, which estimate the poses in the person-centric coordinates, i.e., the coordinates based on the center of the target person. Hence, these methods are inapplicable for multi-person 3D pose estimation, where the absolute coordinates (e.g., the camera coordinates) are required. Moreover, multi-person pose estimation is more challenging than single pose estimation, due to inter-person occlusion and close human interactions. Existing top-down multi-person methods rely on human detection (i.e., top-down approach), and thus suffer from the detection errors and cannot produce reliable pose estimation in multi-person scenes. Meanwhile, existing bottom-up methods that do not use human detection are not affected by detection errors, but since they process all persons in a scene at once, they are prone to errors, particularly for persons in small scales. To address all these challenges, we propose the integration of top-down and bottom-up approaches to exploit their strengths. Our top-down network estimates human joints from all persons instead of one in an image patch, making it robust to possible erroneous bounding boxes. Our bottom-up network incorporates human-detection based normalized heatmaps, allowing the network to be more robust in handling scale variations. Finally, the estimated 3D poses from the top-down and bottom-up networks are fed into our integration network for final 3D poses. To address the common gaps between training and testing data, we do optimization during the test time, by refining the estimated 3D human poses using high-order temporal constraint, re-projection loss, and bone length regularizations. We also introduce a two-person pose discriminator that enforces natural two-person interactions. Finally, we apply a semi-supervised method to overcome the 3D ground-truth data scarcity. Our evaluations demonstrate the effectiveness of the proposed method and its individual components. Our code and pretrained models are available publicly: https://github.com/3dpose/3D-Multi-Person-Pose.

Evaluating current state of monocular 3D pose models for golf

Monocular 3D human pose estimation has reached an impressive performance. State-of-the-art mod- els predict joint locations that can be accurately reprojected back into the image, resulting in vi- sually convincing detections. However, our aim is to use the predicted poses in a domain with high- frequency movements, that is, for video of ath- letes performing golf swings. Our investigation is based on accurate marker-based motion capture data. Also, for our data, the predicted 3D joint locations look convincing when we reproject them into the image. However, by quantitatively com- paring the results with the motion capture data, we see significant model errors that are too erroneous to be used for any kinematic analysis of the move- ments. Thus we conclude that the current models cannot be used out of the box for advanced golf analytics.

Weakly-Supervised 3D Human Pose Estimation With Cross-View U-Shaped Graph Convolutional Network

Although monocular 3D human pose estimation methods have made significant progress, it's far from being solved due to the inherent depth ambiguity. Instead, exploiting multi-view information is a practical way to achieve absolute 3D human pose estimation. In this paper, we propose a simple yet effective pipeline for weakly-supervised cross-view 3D human pose estimation. By only using two camera views, our method can achieve state-of-the-art performance in a weakly-supervised manner, requiring no 3D ground truth but only 2D annotations. Specifically, our method contains two steps: triangulation and refinement. First, given the 2D keypoints that can be obtained through any classic 2D detection methods, triangulation is performed across two views to lift the 2D keypoints into coarse 3D poses. Then, a novel cross-view U-shaped graph convolutional network (CV-UGCN), which can explore the spatial configurations and cross-view correlations, is designed to refine the coarse 3D poses. In particular, the refinement progress is achieved through weakly-supervised learning, in which geometric and structure-aware consistency checks are performed. We evaluate our method on the standard benchmark dataset, Human3.6M. The Mean Per Joint Position Error on the benchmark dataset is 27.4 mm, which outperforms the state-of-the-arts remarkably (27.4 mm vs 30.2 mm).

G2O-Pose: Real-Time Monocular 3D Human Pose Estimation Based on General Graph Optimization.

Monocular 3D human pose estimation is used to calculate a 3D human pose from monocular images or videos. It still faces some challenges due to the lack of depth information. Traditional methods have tried to disambiguate it by building a pose dictionary or using temporal information, but these methods are too slow for real-time application. In this paper, we propose a real-time method named G2O-pose, which has a high running speed without affecting the accuracy so much. In our work, we regard the 3D human pose as a graph, and solve the problem by general graph optimization (G2O) under multiple constraints. The constraints are implemented by algorithms including 3D bone proportion recovery, human orientation classification and reverse joint correction and suppression. When the depth of the human body does not change much, our method outperforms the previous non-deep learning methods in terms of running speed, with only a slight decrease in accuracy.

Staged cascaded network for monocular 3D human pose estimation

Staged cascaded network for monocular 3D human pose estimation

Adapted human pose: monocular 3D human pose estimation with zero real 3D pose data

The ultimate goal for an inference model is to be robust and functional in real life applications. However, training vs. test data domain gaps often negatively affect model performance. This issue is especially critical for the monocular 3D human pose estimation problem, in which 3D human data is often collected in a controlled lab setting. In this paper, we focus on alleviating the negative effect of domain shift in both appearance and pose space for 3D human pose estimation by presenting our adapted human pose (AHuP) approach. AHuP is built upon two key components: (1) semantically aware adaptation (SAA) for the cross-domain feature space adaptation, and (2) skeletal pose adaptation (SPA) for the pose space adaptation which takes only limited information from the target domain. By using zero real 3D human pose data, one of our adapted synthetic models shows comparable performance with the SOTA pose estimation models trained with large scale real 3D human datasets. The proposed SPA can be also employed independently as a light-weighted head to improve existing SOTA models in a novel context. A new 3D scan-based synthetic human dataset called ScanAva+ is also going to be publicly released with this work.

Boosting Monocular 3D Human Pose Estimation With Part Aware Attention.

Monocular 3D human pose estimation is challenging due to depth ambiguity. Convolution-based and Graph-Convolution-based methods have been developed to extract 3D information from temporal cues in motion videos. Typically, in the lifting-based methods, most recent works adopt the transformer to model the temporal relationship of 2D keypoint sequences. These previous works usually consider all the joints of a skeleton as a whole and then calculate the temporal attention based on the overall characteristics of the skeleton. Nevertheless, the human skeleton exhibits obvious part-wise inconsistency of motion patterns. It is therefore more appropriate to consider each part's temporal behaviors separately. To deal with such part-wise motion inconsistency, we propose the Part Aware Temporal Attention module to extract the temporal dependency of each part separately. Moreover, the conventional attention mechanism in 3D pose estimation usually calculates attention within a short time interval. This indicates that only the correlation within the temporal context is considered. Whereas, we find that the part-wise structure of the human skeleton is repeating across different periods, actions, and even subjects. Therefore, the part-wise correlation at a distance can be utilized to further boost 3D pose estimation. We thus propose the Part Aware Dictionary Attention module to calculate the attention for the part-wise features of input in a dictionary, which contains multiple 3D skeletons sampled from the training set. Extensive experimental results show that our proposed part aware attention mechanism helps a transformer-based model to achieve state-of-the-art 3D pose estimation performance on two widely used public datasets. The codes and the trained models are released at https://github.com/thuxyz19/3D-HPE-PAA.

Locally Connected Network for Monocular 3D Human Pose Estimation.

We present an approach for 3D human pose estimation from monocular images. The approach consists of two steps: it first estimates a 2D pose from an image and then estimates the corresponding 3D pose. This paper focuses on the second step. Graph convolutional network (GCN) has recently become the de facto standard for human pose related tasks such as action recognition. However, in this work, we show that GCN has critical limitations when it is used for 3D pose estimation due to the inherent weight sharing scheme. The limitations are clearly exposed through a novel reformulation of GCN, in which both GCN and Fully Connected Network (FCN) are its special cases. In addition, on top of the formulation, we present locally connected network (LCN) to overcome the limitations of GCN by allocating dedicated rather than shared filters for different joints. We jointly train the LCN network with a 2D pose estimator such that it can handle inaccurate 2D poses. We evaluate our approach on two benchmark datasets and observe that LCN outperforms GCN, FCN, and the state-of-the-art methods by a large margin. More importantly, it demonstrates strong cross-dataset generalization ability because of sparse connections among body joints.

Multi-View Pose Generator Based on Deep Learning for Monocular 3D Human Pose Estimation

In this paper, we study the problem of monocular 3D human pose estimation based on deep learning. Due to single view limitations, the monocular human pose estimation cannot avoid the inherent occlusion problem. The common methods use the multi-view based 3D pose estimation method to solve this problem. However, single-view images cannot be used directly in multi-view methods, which greatly limits practical applications. To address the above-mentioned issues, we propose a novel end-to-end 3D pose estimation network for monocular 3D human pose estimation. First, we propose a multi-view pose generator to predict multi-view 2D poses from the 2D poses in a single view. Secondly, we propose a simple but effective data augmentation method for generating multi-view 2D pose annotations, on account of the existing datasets (e.g., Human3.6M, etc.) not containing a large number of 2D pose annotations in different views. Thirdly, we employ graph convolutional network to infer a 3D pose from multi-view 2D poses. From experiments conducted on public datasets, the results have verified the effectiveness of our method. Furthermore, the ablation studies show that our method improved the performance of existing 3D pose estimation networks.

3D human pose estimation with siamese equivariant embedding

In monocular 3D human pose estimation a common setup is to first detect 2D positions and then lift the detection into 3D coordinates. Many algorithms suffer from overfitting to camera positions in the training set. We propose a siamese architecture that learns a rotation equivariant hidden representation to reduce the need for data augmentation. Our method is evaluated on multiple databases with different base networks and shows a consistent improvement of error metrics. It achieves state-of-the-art cross-camera error rate among algorithms that use estimated 2D joint coordinates only.