Human Pose Estimation Model Research Articles

Optimization of ICANet lightweight human pose estimation based on HRNet

The existing high-resolution human pose estimation models have low applicability in practical applications due to their large parameter quantity and high computational complexity. To address these issues, this paper proposes a human pose estimation network with a lightweight inverted residual coordinate attention network (ICANet) based on the high-resolution network (HRNet). With the introduction of CoordAttention mechanism and the inverted residual module, this paper proposes two lightweight network modules, namely ICAneck and ICAblock, which not only reduce model parameter quantity and computational complexity, but also achieve feature enhancement of long-range dependence and precise position information in spatial directions of the feature map. Experimental results show that compared to HRNet, the ICANet model proposed in this paper reduces its parameter quantity by 53.7% and computational complexity by 32.4% on the COCO validation set, and lowers its parameter amount by 53.7% and computational complexity by 32.6% on the MPII validation set. Practical applications prove that ICANet still achieves high-precision detection of human key points with fewer parameters and lower computational complexity, and has higher applicability and practicality compared with common human pose estimation networks such as the Stacked Hourglass Network (Hourglass), Cascaded Pyramid Network (CPN), and SimpleBaseline, and therefore has better applicability and practicality.

An Improved High-Resolution Network-Based Method for Yoga-Pose Estimation

In this paper, SEPAM_HRNet, a high-resolution pose-estimation model that incorporates the squeeze-and-excitation and pixel-attention-mask (SEPAM) module is proposed. Feature pyramid extraction, channel attention, and pixel-attention masks are integrated into the SEPAM module, resulting in improved model performance. The construction of the model involves replacing ordinary convolutions with the plug-and-play SEPAM module, which leads to the creation of the SEPAMneck module and SEPAMblock module. To evaluate the model’s performance, the YOGA2022 human yoga poses teaching dataset is presented. This dataset comprises 15,350 images that capture ten basic yoga pose types—Warrior I Pose, Warrior II Pose, Bridge Pose, Downward Dog Pose, Flat Pose, Inclined Plank Pose, Seated Pose, Triangle Pose, Phantom Chair Pose, and Goddess Pose—with a total of five participants. The YOGA2022 dataset serves as a benchmark for evaluating the accuracy of the human pose-estimation model. The experimental results demonstrated that the SEPAM_HRNet model achieved improved accuracy in predicting human keypoints on both the common objects in context (COCO) calibration set and the YOGA2022 calibration set, compared to other state-of-the-art human pose-estimation models with the same image resolution and environment configuration. These findings emphasize the superior performance of the SEPAM_HRNet model.

Distributed Human Pose Estimation Based on BlazePose

The lightweight of human pose estimation model is one current research hotspot in the field of computer vision. The emphasis of current research is to promise lightweight model while improving the network performance. In view of computing and processing performance of monocular static image pose estimation method deployed in mobile devices, a distributed human pose estimation method was proposed. The distributed framework was constructed by simplifying BlazePose as the core, and the 5×5 convolution kernels were run for feature downsampling in a multi-thread simulated distributed environment. The computing power output of mobile devices was reduced by distributed simulation. Experimental results showed that the proposed model had 3%~11% performance improvement compared with other models in the simulation environment. Compared with the traditional human pose estimation model, the proposed model reconstructs the neural network deployment framework, thus indirectly enabling stable recognition of the deployed equipment under low computing power.

Ergonomic investigations on novel dynamic postural estimator using blaze pose and transfer learning

The aim is to develop a computer-based assessment model for novel dynamic postural evaluation using RULA. The present study proposed a camera-based, three-dimensional (3D) dynamic human pose estimation model using ‘BlazePose’ with a data set of 50,000 action-level-based images. The model was investigated using the Deep Neural Network (DNN) and Transfer Learning (TL) approach. The model has been trained to evaluate the posture with high accuracy, precision, and recall for each output prediction class. The model can quickly analyse the ergonomics of dynamic posture online and offline with a promising accuracy of 94.12%. A novel dynamic postural estimator using blaze pose and transfer learning is proposed and assessed for accuracy. The model is subjected to a constant muscle loading factor and foot support score that could evaluate one person with good image clarity at a time. Practitioner summary: A detailed investigation of dynamic work postures is largely missing in the literature. Experimental analysis has been performed using transfer learning, BlazePose, and RULA action levels. An overall accuracy of 94.12% is achieved for dynamic postural assessment.

HPnet: Hybrid Parallel Network for Human Pose Estimation.

Hybrid models which combine the convolution and transformer model achieve impressive performance on human pose estimation. However, the existing hybrid models on human pose estimation, which typically stack self-attention modules after convolution, are prone to mutual conflict. The mutual conflict enforces one type of module to dominate over these hybrid sequential models. Consequently, the performance of higher-precision keypoints localization is not consistent with overall performance. To alleviate this mutual conflict, we developed a hybrid parallel network by parallelizing the self-attention modules and the convolution modules, which conduce to leverage the complementary capabilities effectively. The parallel network ensures that the self-attention branch tends to model the long-range dependency to enhance the semantic representation, whereas the local sensitivity of the convolution branch contributes to high-precision localization simultaneously. To further mitigate the conflict, we proposed a cross-branches attention module to gate the features generated by both branches along the channel dimension. The hybrid parallel network achieves 75.6% and 75.4%AP on COCO validation and test-dev sets and achieves consistent performance on both higher-precision localization and overall performance. The experiments show that our hybrid parallel network is on par with the state-of-the-art human pose estimation models.

Human pose estimation model based on DiracNets and integral pose regression

Human pose estimation model based on DiracNets and integral pose regression

PENGIRAAN POSE MODEL MANUSIA PADA REPETISI KEBUGARAN AI PEMOGRAMAN PYTHON BERBASIS KOMPUTERISASI

Pandemic people don't have free access to Gym. The importance of keeping the body fit. This can be overcome with the Human Pose Estimation Computer Vision technique. Optimum and maximum distance testing data. Webcam detects movement and test data on AI Fitness Counter models. Human Pose Estimation detection data is taken when someone does fitness exercises such as pull ups, pushups, and lifting weights. Optimal webcam detects Human Pose Estimation model AI Fitness Counter is three meters. Every detail of the landmarks in the pose corresponds to the key points of each limb and the utility lines that connect form the skeleton of the body. The application of the number of detection repetitions takes advantage of the angle of the elbow when it is straight and when it is bent. The angle forms 300 then a stage up occurs and is counted as one repetition. Meanwhile, when the angle forms 1700, a stage down occurs, and one repetition is counted as well. The use of Media pipe for the detection results is accurate and effective. This detection was successful because all members of the body were detected.

Performance Analysis of Human Activity

This project aims to develop an AI-powered gym assistant using Jupyter Notebook and MediaPipe, a popular computer vision library, to count the repetitions of three joint exercises: curls, squats, and sit-ups. The system will provide real-time feedback and monitoring, allowing users to track their progress and improve performance. The proposed method utilizes MediaPipe, which offers pre-trained machine-learning models for human pose estimation and hand tracking. These models will accurately detect and track critical body joints and hand movements during the exercises. The system will then analyze the detected poses to identify the repetitions of each activity based on predefined movement patterns and pose thresholds. Jupyter Notebook will be used as the development environment for coding and testing the system. Python programming language and MediaPipe’s Python API will be employed to implement the pose estimation and repetition counting algorithms. The system will also incorporate a user-friendly interface, allowing users to interact with the gym assistant and receive feedback on their exercise performance. The completed project will provide an AI-powered gym assistant that can accurately count the repetitions of curls, squats, and sit-ups in real-time. Additionally, this project will contribute to the advancement of the field of fitness technology by showcasing the potential of combining computer vision and artificial intelligence techniques for gym monitoring and performance tracking. The results of this project have the potential to benefit fitness enthusiasts, trainers, and researchers alike, providing contributions to the field of fitness technology.

Detecting and grouping keypoints for multi-person pose estimation using instance-aware attention

Bottom-up human pose estimation models detect keypoints and learn associative information between keypoints, usually requiring human predefined offset fields or embeddings for keypoints grouping (clustering). In this paper, we present a brand new method that can entirely solve these problems based on Transformer, making the grouping process free of the human-defined associative signals. Specifically, the self-attention in vision Transformer measures feature similarity between any pair of locations, which provides a metric space to associate keypoints together into corresponding human instances. However, the naive attention patterns formed in Transformer are still not subjectively controlled, so there is no guarantee that the keypoints only attend to the instances to which they belong. To address it we propose a novel approach of supervising self-attention to be instance-aware, simultaneously accomplishing multi-person keypoint detection and clustering. By doing so, we can group the detected keypoints to their corresponding instances, according to the pairwise attention scores.An additional benefit of our method is that the instance segmentation results of any number of people can be directly obtained from the supervised attention matrix, thereby simplifying the pixel assignment pipeline. The qualitative and quantitative results on the COCO shows that, with a very simple architecture design, our method can achieve comparable performance against the CNN-based bottom-up counterparts with fewer parameters, which also demonstrate a promising way to control self-attention mechanism behavior for specific purposes.

2D Human pose estimation: a survey

Human pose estimation aims at localizing human anatomical keypoints or body parts in the input data (e.g., images, videos, or signals). It forms a crucial component in enabling machines to have an insightful understanding of the behaviors of humans, and has become a salient problem in computer vision and related fields. Deep learning techniques allow learning feature representations directly from the data, significantly pushing the performance boundary of human pose estimation. In this paper, we reap the recent achievements of 2D human pose estimation methods and present a comprehensive survey. Briefly, existing approaches put their efforts in three directions, namely network architecture design, network training refinement, and post processing. Network architecture design looks at the architecture of human pose estimation models, extracting more robust features for keypoint recognition and localization. Network training refinement tap into the training of neural networks and aims to improve the representational ability of models. Post processing further incorporates model-agnostic polishing strategies to improve the performance of keypoint detection. More than 200 research contributions are involved in this survey, covering methodological frameworks, common benchmark datasets, evaluation metrics, and performance comparisons. We seek to provide researchers with a more comprehensive and systematic review on human pose estimation, allowing them to acquire a grand panorama and better identify future directions.

HQLC-Overlap: an adaptive low-cost binocular 3D human pose estimation model

HQLC-Overlap: an adaptive low-cost binocular 3D human pose estimation model

Pose estimation method for construction machine based on improved AlphaPose model

PurposeThe safety management of construction machines is of primary importance. Considering that traditional construction machine safety monitoring and evaluation methods cannot adapt to the complex construction environment, and the monitoring methods based on sensor equipment cost too much. This paper aims to introduce computer vision and deep learning technologies to propose the YOLOv5-FastPose (YFP) model to realize the pose estimation of construction machines by improving the AlphaPose human pose model.Design/methodology/approachThis model introduced the object detection module YOLOv5m to improve the recognition accuracy for detecting construction machines. Meanwhile, to better capture the pose characteristics, the FastPose network optimized feature extraction was introduced into the Single-Machine Pose Estimation Module (SMPE) of AlphaPose. This study used Alberta Construction Image Dataset (ACID) and Construction Equipment Poses Dataset (CEPD) to establish the dataset of object detection and pose estimation of construction machines through data augmentation technology and Labelme image annotation software for training and testing the YFP model.FindingsThe experimental results show that the improved model YFP achieves an average normalization error (NE) of 12.94 × 10–3, an average Percentage of Correct Keypoints (PCK) of 98.48% and an average Area Under the PCK Curve (AUC) of 37.50 × 10–3. Compared with existing methods, this model has higher accuracy in the pose estimation of the construction machine.Originality/valueThis study extends and optimizes the human pose estimation model AlphaPose to make it suitable for construction machines, improving the performance of pose estimation for construction machines.

Jersey number detection using synthetic data in a low-data regime.

Player identification is an essential and complex task in sports video analysis. Different strategies have been devised over the years and identification based on jersey numbers is one of the most common approaches given its versatility and relative simplicity. However, automatic detection of jersey numbers is challenging due to changing camera angles, low video resolution, small object size in wide-range shots, and transient changes in the player's posture and movement. In this paper, we present a novel approach for jersey number identification in a small, highly imbalanced dataset from the Seattle Seahawks practice videos. We generate novel synthetic datasets of different complexities to mitigate the data imbalance and scarcity in the samples. To show the effectiveness of our synthetic data generation, we use a multi-step strategy that enforces attention to a particular region of interest (player's torso), to identify jersey numbers. The solution first identifies and crops players in a frame using a person detection model, then utilizes a human pose estimation model to localize jersey numbers in the detected players, obviating the need for annotating bounding boxes for number detection. We experimented with two sets of Convolutional Neural Networks (CNNs) with different learning objectives: multi-class for two-digit number identification and multi-label for digit-wise detection to compare performance. Our experiments indicate that our novel synthetic data generation method improves the accuracy of various CNN models by 9% overall, and 18% on low frequency numbers.

Validity of an artificial intelligence, human pose estimation model for measuring single-leg squat kinematics

Few studies have investigated the validity of 2D pose estimation models to evaluate kinematics throughout a motion and none have included adolescents. Adolescent athletes completed single-leg squats while 3D kinematic data and 2D sagittal and frontal plane videos were recorded. Sagittal and frontal plane joint motion throughout the single-leg squats and angles at peak knee flexion were compared among 2D pose estimation, 3D motion analysis, and traditional 2D motion analysis techniques. Statistical parametric mapping compared waveforms while Pearson’s correlation compared the relationships of joint angles at peak knee flexion among techniques, respectively. We observed significant waveform differences between 2D pose estimation and 3D motion analysis at the beginning and end of the squat for sagittal plane hip and knee motion, for most of the squat for frontal plane hip motion, and throughout the entire squat for sagittal plane ankle motion and frontal plane pelvic motion. We observed moderate-to-strong relationships (r = 0.68-0.94) for sagittal plane joint angles between 2D pose estimation and 3D techniques. We observed fair correlations (r = 0.53-0.54) for frontal plane pelvic and hip joint angles between 2D pose estimation and 3D motion analysis. We observed poor relationships for the frontal plane knee angle between 3D motion analysis with 2D pose estimation (r = 0.20) and traditional 2D motion analysis (r = 0.05), respectively, but observed a strong relationship (r = 0.95) between the 2D techniques. 2D pose estimation is a valid alternative to 3D motion analysis and traditional 2D motion analysis for evaluating most sagittal and frontal plane angles during a single-leg squatting task.

Synthesising 2D Video from 3D Motion Data for Machine Learning Applications.

To increase the utility of legacy, gold-standard, three-dimensional (3D) motion capture datasets for computer vision-based machine learning applications, this study proposed and validated a method to synthesise two-dimensional (2D) video image frames from historic 3D motion data. We applied the video-based human pose estimation model OpenPose to real (in situ) and synthesised 2D videos and compared anatomical landmark keypoint outputs, with trivial observed differences (2.11–3.49 mm). We further demonstrated the utility of the method in a downstream machine learning use-case in which we trained and then tested the validity of an artificial neural network (ANN) to estimate ground reaction forces (GRFs) using synthesised and real 2D videos. Training an ANN to estimate GRFs using eight OpenPose keypoints derived from synthesised 2D videos resulted in accurate waveform GRF estimations (r > 0.9; nRMSE < 14%). When compared with using the smaller number of real videos only, accuracy was improved by adding the synthetic views and enlarging the dataset. The results highlight the utility of the developed approach to enlarge small 2D video datasets, or to create 2D video images to accompany 3D motion capture datasets to make them accessible for machine learning applications.

Human action recognition using a convolutional neural network based on skeleton heatmaps from two-stage pose estimation

Human action recognition based on skeleton information has been extensively used in various areas, such as human–computer interaction. In this paper, we extracted human skeleton data by constructing a two-stage human pose estimation model, which combined the improved single shot detector (SSD) algorithm with convolutional pose machines (CPM) to obtain human skeleton heatmaps. The backbone of the SSD algorithm was replaced with ResNet, which can characterize images effectively. In addition, we designed multiscale transformation rules for CPM to fuse the information of different scales and a convolutional neural network for the classification of the skeleton keypoints heatmaps to complete action recognition. Indoor and outdoor experiments were conducted on the Caster Moma mobile robot platform, and without an external remote control, the real-time movement of the robot was controlled by the leader through command actions.

Deep Learning Networks for View-independent Knee and Elbow Joint Angle Estimation.

Vision-based human joint angle estimation is essential for remote and continuous health monitoring. Most vision-based angle estimation methods use the locations of human joints extracted using optical motion cameras, depth cameras, or human pose estimation models. This study aimed to propose a reliable and straightforward approach with deep learning networks for knee and elbow flexion/extension angle estimation from the RGB video. Fifteen healthy participants performed four daily activities in this study. The experiments were conducted with four different deep learning networks, and the networks took nine subsequent frames as input while output was knee and elbow joint angles extracted from an optical motion capture system for each frame. The BiLSTM network-based joint angles estimator can estimate both joint angles with a correlation of 0.955 for knee and 0.917 for elbow joints regardless of the camera view angles.

Application of Deep Convolution Network Algorithm in Sports Video Hot Spot Detection.

Sports videos are blowing up over the internet with enriching material life and the higher pursuit of spiritual life of people. Thus, automatically identifying and detecting helpful information from videos have arisen as a relatively novel research direction. Accordingly, the present work proposes a Human Pose Estimation (HPE) model to automatically classify sports videos and detect hot spots in videos to solve the deficiency of traditional algorithms. Firstly, Deep Learning (DL) is introduced. Then, amounts of human motion features are extracted by the Region Proposal Network (RPN). Next, an HPE model is implemented based on Deep Convolutional Neural Network (DCNN). Finally, the HPE model is applied to motion recognition and video classification in sports videos. The research findings corroborate that an effective and accurate HPE model can be implemented using the DCNN to recognize and classify videos effectively. Meanwhile, Big Data Technology (BDT) is applied to count the playing amounts of various sports videos. It is convinced that the HPE model based on DCNN can effectively and accurately classify the sports videos and then provide a basis for the following statistics of various sports videos by BDT. Finally, a new outlook is proposed to apply new technology in the entertainment industry.

Lightweight and Effective Human Pose Estimation Model Based on Multi-Angle Knowledge Distillation

In the field of human pose estimation, most of the existing methods focus on improving the generalization performance of the model, while ignoring the significant efficiency issues. This leads to an increasing amount of model parameters and needs to take up more and more computing resources, which greatly reduces the practical value of the model. In order to solve this problem, we propose a novel lightweight network structure called Effective and Lightweight Pose Network (ELPN). At the same time, for the sake of alleviating the difficulty of lightweight model training, we propose a Multi-Angle Pose Distillation (MAPD) model training method that can more effectively train particularly small pose network models. In quantitative experiments, our models have excellent performance on two mainstream benchmark datasets: the MPII and the COCO. In qualitative testing, our models can accurately locate the keypoints of complex human movements. These fully demonstrates the efficiency and effectiveness of our methods. Our models have the characteristics of high precision, small size and fast inference speed. It is a cost-effective model with greater practical value.

LHPE-nets: A lightweight 2D and 3D human pose estimation model with well-structural deep networks and multi-view pose sample simplification method.

The cross-view 3D human pose estimation model has made significant progress, it better completed the task of human joint positioning and skeleton modeling in 3D through multi-view fusion method. The multi-view 2D pose estimation part of this model is very important, but its training cost is also very high. It uses some deep learning networks to generate heatmaps for each view. Therefore, in this article, we tested some new deep learning networks for pose estimation tasks. These deep networks include Mobilenetv2, Mobilenetv3, Efficientnetv2 and Resnet. Then, based on the performance and drawbacks of these networks, we built multiple deep learning networks with better performance. We call our network in this article LHPE-nets, which mainly includes Low-Span network and RDNS network. LHPE-nets uses a network structure with evenly distributed channels, inverted residuals, external residual blocks and a framework for processing small-resolution samples to achieve training saturation faster. And we also designed a static pose sample simplification method for 3D pose data. It implemented low-cost sample storage, and it was also convenient for models to read these samples. In the experiment, we used several recent models and two public estimation indicators. The experimental results show the superiority of this work in fast start-up and network lightweight, it is about 1-5 epochs faster than the Resnet-34 during training. And they also show the accuracy improvement of this work in estimating different joints, the estimated performance of approximately 60% of the joints is improved. Its performance in the overall human pose estimation exceeds other networks by more than 7mm. The experiment analyzes the network size, fast start-up and the performance in 2D and 3D pose estimation of the model in this paper in detail. Compared with other pose estimation models, its performance has also reached a higher level of application.