Regions Of Human Body Research Articles

Real-time pose estimation and motion tracking for motion performance using deep learning models

Abstract With the refinement and scientificization of sports training, the demand for sports performance analysis in the field of sports has gradually become prominent. In response to the problem of low accuracy and poor real-time performance in human pose estimation during sports, this article focused on volleyball sports and used a combination model of OpenPose and DeepSORT to perform real-time pose estimation and tracking on volleyball videos. First, the OpenPose algorithm was adopted to estimate the posture of the human body region, accurately estimating the coordinates of key points, and assisting the model in understanding the posture. Then, the DeepSORT model target tracking algorithm was utilized to track the detected human pose information in real-time, ensuring consistency of identification and continuity of position between different frames. Finally, using unmanned aerial vehicles as carriers, the YOLOv4 object detection model was used to perform real-time human pose detection on standardized images. The experimental results on the Volleyball Activity Dataset showed that the OpenPose model had a pose estimation accuracy of 98.23%, which was 6.17% higher than the PoseNet model. The overall processing speed reached 16.7 frames/s. It has good pose recognition accuracy and real-time performance and can adapt to various volleyball match scenes.

Discomfort survey of dumper operators exposed to whole-body vibration in opencast coal mines

Exposure to whole-body vibration (WBV) can lead to pain and discomfort of the heavy earthmoving machinery operators. In this paper, WBV measurement of 110 dumper operators has been conducted in three opencast coal mines, followed by a discomfort survey. Vibration measurement revealed that the mean frequency-weighted daily rms acceleration, A(8), for the operators was 0.87 m s−2. The discomfort survey was conducted to determine whether body regions of discomfort relate to their kinds of task performed by the dumper operators and also to prioritize areas for ergonomic intervention. Overall discomfort was evaluated corresponding to five regions of human body in the present survey. The results of the survey revealed that maximum operators had discomfort in their lower back region (65%) followed by leg region (53%). The industrial application of discomfort analysis is to identify the human body regions susceptible to vibration hazards. The discomfort indices are also correlated with vibration magnitude and some of the contributing factors of WBV for better understanding the vibration risks faced by the dumper operators for effectively managing the impacts of vibration in surface coal mines.

Tomographic Magnetic Particle Imaging with a Single-Sided Field-Free Line Scanner.

Magnetic Particle Imaging (MPI) is a biomedical imaging modality that shows promise in enhancing clinical imaging capabilities. While there are current efforts to expand MPI hardware to enable wholebody and head imaging, this comes at a significant cost in terms of complexity and expense. A single-sided scanner can offer greater accessibility to the scanning area, as all hardware is located on one side of the device's surface, however, at the price of the limited penetration depth. Despite this, a single-sided device could serve as an open geometry preclinical scanner and a clinical instrument for local screening and procedures. The original single-sided device was introduced for a field-free point encoding gradient, which may limit its practical characteristics. We developed a field-free line single-sided device that has improved imaging characteristics, which is beneficial for preclinical applications and potential clinical translation. We demonstrated full tomographic imaging capabilities with the prototype of a single-sided MPI scanner and characterized the imaging performance of our imaging system. MPI has the potential to image various human body regions, particularly for breast cancer diagnosis. Our research tackles the crucial aspect of scalability in the emerging MPI technique.

Multi-supervision transformer combining bounding box and mask for data-limited pose estimation

In scenarios where acquiring accurate annotated pose data is difficult, obtaining sufficient training samples becomes a challenge. Given these data constraints, it is challenging to provide optimal end-to-end supervision with limited pose information. To address this limitation, we propose a novel approach, named MSPose, which combines diverse data sources from human instances to provide multi-level supervision in human pose estimation. This approach mitigates the impact of scarce training data and enhances pose estimation precision by synergistically incorporating human bounding box and mask information. Our proposed method achieves multi-level model activation for the human body region, structure, and joints. This hierarchical pixel activation is facilitated by integrating human bounding box and mask signals as supplementary supervision during training. Importantly, we exclude auxiliary branches to ensure complete freedom during inference. Our experimental validation on a quarter of the COCO and MPII datasets demonstrates the competitiveness of our method compared to state-of-the-art methods. Compared to TokenPose-T, our MSPose-T achieves a 6.1 points improvement in average precision and a corresponding 3.1 points increase on the full COCO validation set. On the MPII dataset, MSPose achieves the highest Mean score of 1.9 points. Our approach achieves these improvements while maintaining consistent parameters and GFLOPs, highlighting the advantage of our approach.

Generalisation of radiotherapy dose calculation for Monte Carlo algorithm combined with 3D Swin-Unet: a multi-institutional IMRT evaluation

Objective. Accurate dose calculations are essential prerequisites for precise radiotherapy. The integration of deep learning into dosimetry could consider computational accuracy and efficiency and has potential applicability to clinical dose calculation. The generalisation of a deep learning dose calculation method (hereinafter referred to as TERMA-Monte Carlo network, T-MC net) was evaluated in clinical practice using intensity-modulated radiotherapy (IMRT) plans for various human body regions and multiple institutions, with the Monte Carlo (MC) algorithm serving as a benchmark. Approach. Sixty IMRT plans were selected from four institutions for testing the head and neck, chest and abdomen, and pelvis regions. Using the MC results as the benchmark, the T-MC net calculation results were used to perform three-dimensional dose distribution and dose-volume histogram (DVH) comparisons of the entire body, planning target volume (PTV) and organs at risk (OARs), respectively, and calculate the mean ±95% confidence interval of gamma pass rate (GPR), percentage of agreement (PA) and dose difference ratio (DDR) of dose indices D95, D50, and D5. Main results. For the entire body, the GPRs of 3%/3 mm, 2%/2 mm, 2%/1 mm, and the PA were 99.62 ± 0.32%, 98.50 ± 1.09%, 95.60 ± 2.90% and 97.80 ± 1.12%, respectively. For the PTV, the GPRs of 3%/3 mm, 2%/2 mm, 2%/1 mm and the PA were 98.90 ± 1.00%, 95.78 ± 2.83%, 92.23 ± 4.74% and 98.93 ± 0.62%, respectively. The absolute value of average DDR was less than 1.4%. Significance. We proposed a general dose calculation framework based on deep learning, using the MC algorithm as a benchmark, performing a generalisation test for IMRT treatment plans across multiple institutions. The framework provides high computational speed while maintaining the accuracy of MC and may become an effective dose algorithm engine in treatment planning, adaptive radiotherapy, and dose verification.

Quantifying injury risks of collaborative robots in a manufacturing cell using a simulation approach

Safety remains an important impediment towards fully exploiting the potential of collaborative robots for executing manufacturing tasks. The ISO 15066 proposes requirements for safety assurance of collaborative robots. This includes safeguards such as force, torque, and safe distance limiting functions. More importantly, safe force (and impact pressure) limits depend on the human body region. The latter is based on biomechanical limits derived from the body model, which specifies maximum pressure and force limiting values on different body regions. However, the limiting values proposed in the body model seldom considers practical impact scenarios that may potentially occur while executing collaborative manufacturing tasks. For instance, the impact force/pressure thresholds are influenced by aspects such as the speed of the moving manipulator, or contact point, further influenced by the mass, and contact surface area of the end effector. Consequently, this influences operator safety and design of collaborative workspaces and manufacturing tasks. So far, these practical influences are under-explored in literature, discussing human-robot collaboration, while considering manufacturing tasks.In this paper, a simulation framework is proposed exploring a use case of assembly tasks in a manufacturing setting. The framework embeds a hazard and operability analysis to identify potential hazardous events during execution of the collaborative manufacturing tasks. The tasks further modelled in the ABB RobotStudio environment, where varying path configurations of the manipulator are mimicked during task execution. Force and pressure impact values based on the path configurations are quantified on different body regions. Transient and quasi static forces/pressure are considered for varying impact scenarios on the body regions. Finally, risk mitigation is proposed based on the impact scenarios and quantified injury severity. This includes task re-design or implementing alternative path configurations for the robotic manipulator.

A new imbalance-aware loss function to be used in a deep neural network for colorectal polyp segmentation

Colorectal cancers may occur in colon region of human body because of late detection of polyps. Therefore, colonoscopists often use colonoscopy device to view the entire colon in their routine practice to remove polyps by excisional biopsy. The aim of this study is to develop a new imbalance-aware loss function, i.e., omni-comprehensive loss, to be used in deep neural networks to overcome both imbalanced dataset and the vanishing gradient problem in identifying the related regions of a polyp. Another reason of developing a new loss function is to be able to produce a more comprehensive one that has evaluation capabilities of region-based, shape-aware, and pixel-wise distribution loss approaches at once. To measure the performance of the new loss function, two scenarios have been conducted. First, an 18-layer residual network as backbone with UNet as the decoder is implemented. Second, a 34-layer residual network as the encoder and a UNet as the decoder is designed. For both scenarios, the results of using popular imbalance-aware losses are compared with those of using our proposed new loss function. During training and 5-fold cross validation steps, multiple publicly available datasets are used. In addition to original data in these datasets, their augmented versions are also created by flipping, scaling, rotating and contrast-limited adaptive histogram equalization operations. As a result, our proposed new custom loss function produced the best performance metrics compared with the popular loss functions.

PsLSNetV2: End to end deep learning system for measurement of area score of psoriasis regions in color images

ObjectiveAt present, psoriasis area scores are measured manually by dermatologists through visual observations. This subjective method suffers from numerous typical problems. The only solution to these problems is to design and implement objective methods for this. However, most of the existing works in this regard are based on machine learning frameworks that are semi-automated and feature-dependent. In this work, a deep learning-based fully automated, and single-stage framework is proposed to detect psoriasis lesions and measure their area score from color images of human body regions. MethodsThe proposed method is an extension of the existing PsLSNet proposed by our team, which provides a fully automated approach for the segmentation of single psoriasis lesions from cropped patches of skin images. For this proposed work, a new version PsLSNetV2 model is developed for automated segmentation of healthy skin, multiple psoriasis lesions, and background region simultaneously in complete body region images. This proposed model utilizes an efficient and lightweight network with transfer learning to increase the representational efficiency for multi-class segmentation. ResultsThe proposed model is tested by 5-fold cross-validation on a self-generated dataset having 500 images from 100 psoriasis patients. The multi-class segmentation performance of the proposed model achieves an overall Dice-Coefficient Index and Jaccard Index of 97.43% and 95.05% respectively and outperforms the existing models. ConclusionThe fully automated multi-class segmentation results by the proposed lightweight segmentation model are promising enough to determine psoriasis area score objectively with an average accuracy of 94.20% for assisting dermatologists in a simple and rapid way.

Stage Performance Characteristics of Minority Dance Based on Human Motion Recognition

Due to the different development history of different nationalities, there is a big gap between traditional cultures. Especially, as an important embodiment and component of traditional culture, ethnic minority dance has unique characteristics. With the development of machine vision technology, human motion recognition has gradually become a hot research direction, but its application research in the field of dance motion recognition is still in its infancy. In this paper, the characteristics of stage performances of ethnic minority dances in the field of intelligent auxiliary training based on human motion recognition technology are introduced. The human body regions in the frame are obtained by using human posture, and 3D_SIFT and optical flow features are extracted from each region. Then, we use the extracted key frames and DTW (dynamic time warping) algorithm to realize the motion recognition of the motion capture data and carry out simulation experiments. The test results show that the algorithm can identify the minority dance video database, effectively identify the dance movements, and realize the movement correction of dancers.

Attribute-Guided Global and Part-Level Identity Network for Person Re-Identification

Most of the person re-identification (re-ID) algorithms based on deep learning mainly learn the global feature representation of pedestrians, while ignoring the important role of fine-grained pedestrian attribute features on re-ID tasks. Pedestrian attributes are middle-level semantic features, which have invariance in different poses, camera views, and illumination conditions. Considering the robustness and promotion of pedestrian attributes for person re-ID task, we propose an Attribute-guided Global and Part-level identity Network (AGPNet), which consists of a global identity task, a part-level identity task, and a pedestrian attributes learning task. AGPNet takes advantage of perceived semantic information of pedestrian attributes and deploys them as guidance to attend to human body regions and learn robust feature representation in the feature representation construction stage. Extensive experiments on two large-scale person re-ID datasets (Market-1501 and DukeMTMC-reID) show the effectiveness of our method, which is competitive with the state-of-the-art algorithms.

Localization and tracking of closely-spaced human targets based on infrared sensors

In recent years, indoor human localization and tracking technology has been a research hotspot. However, the problem caused by closely-spaced targets is often avoided for target localization and tracking. As a common situation in real scene, it is of great research value and application prospect to find out ways to effectively solve the problem of accurately locating and tracking closely-spaced targets. Thermopile array (TPA) sensors, a kind of widely investigated infrared sensors, are often used to detect human targets, which are expected to solve the closely-spaced problem. This paper proposes a method for localization and tracking of indoor closely-spaced targets by using TPA sensors. Visual background extractor (ViBE) algorithm is used to preprocess the data and extract the infrared characteristic region of human body. Several algorithms such as connected components labelling algorithm, temperature peak method and k-means++ clustering algorithm are used to locate the closely-spaced targets. In the aspect of closely-spaced target tracking algorithm, the temperature summation of target’s infrared thermal map is used to assist the nearest neighbor data association algorithm for data association of closely-spaced targets. Kuhn-Munkres algorithm is used to match the location point and trajectory properly. The Kalman filter, the first-order particle filter and the second-order particle filter are compared and analyzed to further reduce the tracking error. Finally, the system is tested in the actual indoor scene. The experimental results show that the proposed system can achieve the goal of localization and tracking of closely-spaced targets.

Evaluating the Repeatability of Friction Coefficient Measurements and Tactile Perceptions in Skin\u2013Textile Interactions Across Body Regions

Measuring the friction between human skin and textiles is essential to preventing skin-related health problems and wearer discomfort. This study sought to evaluate the repeatability of friction coefficient measurements and tactile perceptions at various human body regions, using a recently developed portable friction measuring device. Using the device, a cotton textile was applied to six test locations. The friction coefficient and tactile perceptions were taken three times, at each test site. The repeatability of friction coefficient was excellent with ICC of 0.91 ± 0.05 for static and 0.91 ± 0.06 for dynamic friction coefficient, and tactile perception showed good to excellent repeatability. The friction coefficient and tactile perception significantly differed across the body regions (p = 0.008 and p = 0.003, respectively). The chest showed the highest static friction coefficient with 0.90 ± 0.19 and dynamic friction coefficient with 0.79 ± 0.20, whereas the dorsal forearm with static 0.31 ± 0.07 and dynamic 0.25 ± 0.05 was the lowest. The ratings of tactile perceptions were independent of friction coefficient when the changes in the coefficient of friction were induced by regional difference, and/or the range of the coefficient of friction values were small. Moreover, skin temperature and friction coefficient correlated positively, whilst cutaneous hydration and friction coefficient did not.

Three-dimensional printed orthosis in biomedical application: A short review

The three-dimensional (3D) printing in medical implants unlocks unparalleled opportunities to completely configure the product to the patient’s measurements and needs. To be noted, the use of personalized 3D printed orthosis used in regeneration for serious orthosis implants of specific patients is growing to date. The 3D printed is unique to the patient instruments that can be used to facilitate correct positioning of implants and improved functional outcomes. The 3D printing, also defined as ‘rapid prototyping’ and ‘additive manufacturing’ is widely regarded as the ‘second technological revolution. The orthosis is an “externally applied mechanism used to alter the structural and functional properties of the musculoskeletal and skeletal system”. Applications in orthosis healthcare that are pioneering the way 3D printing is performed, changing the orthosis implant markets. This paper is reporting literature on the development of orthosis using 3D printing technology that could make the users more comfortable and easier to maintain. From the literature search, this paper summarises some important information about the use of 3D printing for orthosis development where it focusses on specific regions of human body, the materials for the 3D printed orthosis and further directions of this technology and research. In conclusion, the findings from this review paper may lead to a future recommendation and study in providing better treatment for patients.

KGSNet: Key-Point-Guided Super-Resolution Network for Pedestrian Detection in the Wild.

In real-world scenarios (i.e., in the wild), pedestrians are often far from the camera (i.e., small scale), and they often gather together and occlude with each other (i.e., heavily occluded). However, detecting these small-scale and heavily occluded pedestrians remains a challenging problem for the existing pedestrian detection methods. We argue that these problems arise because of two factors: 1) insufficient resolution of feature maps for handling small-scale pedestrians and 2) lack of an effective strategy for extracting body part information that can directly deal with occlusion. To solve the above-mentioned problems, in this article, we propose a key-point-guided super-resolution network (coined KGSNet) for detecting these small-scale and heavily occluded pedestrians in the wild. Specifically, to address factor 1), a super-resolution network is first trained to generate a clear super-resolution pedestrian image from a small-scale one. In the super-resolution network, we exploit key points of the human body to guide the super-resolution network to recover fine details of the human body region for easier pedestrian detection. To address factor 2), a part estimation module is proposed to encode the semantic information of different human body parts where four semantic body parts (i.e., head and upper/middle/bottom body) are extracted based on the key points. Finally, based on the generated clear super-resolved pedestrian patches padded with the extracted semantic body part images at the image level, a classification network is trained to further distinguish pedestrians/backgrounds from the inputted proposal regions. Both proposed networks (i.e., super-resolution network and classification network) are optimized in an alternating manner and trained in an end-to-end fashion. Extensive experiments on the challenging CityPersons data set demonstrate the effectiveness of the proposed method, which achieves superior performance over previous state-of-the-art methods, especially for those small-scale and heavily occluded instances. Beyond this, we also achieve state-of-the-art performance (i.e., 3.89% MR-2 on the reasonable subset) on the Caltech data set.

Pointless Pose: Part Affinity Field-Based 3D Pose Estimation without Detecting Keypoints

Human pose estimation finds its application in an extremely wide domain and is therefore never pointless. We propose in this paper a new approach that, unlike any prior one that we are aware of, bypasses the 2D keypoint detection step based on which the 3D pose is estimated, and is thus pointless. Our motivation is rather straightforward: 2D keypoint detection is vulnerable to occlusions and out-of-image absences, in which case the 2D errors propagate to 3D recovery and deteriorate the results. To this end, we resort to explicitly estimating the human body regions of interest (ROI) and their 3D orientations. Even if a portion of the human body, like the lower arm, is partially absent, the predicted orientation vector pointing from the upper arm will take advantage of the local image evidence and recover the 3D pose. This is achieved, specifically, by deforming a skeleton-shaped puppet template to fit the estimated orientation vectors. Despite its simple nature, the proposed approach yields truly robust and state-of-the-art results on several benchmarks and in-the-wild data.

Incremental Generative Occlusion Adversarial Suppression Network for Person ReID.

Person re-identification (re-id) suffers from the significant challenge of occlusion, where an image contains occlusions and less discriminative pedestrian information. However, certain work consistently attempts to design complex modules to capture implicit information (including human pose landmarks, mask maps, and spatial information). The network, consequently, focuses on discriminative features learning on human non-occluded body regions and realizes effective matching under spatial misalignment. Few studies have focused on data augmentation, given that existing single-based data augmentation methods bring limited performance improvement. To address the occlusion problem, we propose a novel Incremental Generative Occlusion Adversarial Suppression (IGOAS) network. It consists of 1) an incremental generative occlusion block, generating easy-to-hard occlusion data, that makes the network more robust to occlusion by gradually learning harder occlusion instead of hardest occlusion directly. And 2) a global-adversarial suppression (G&A) framework with a global branch and an adversarial suppression branch. The global branch extracts steady global features of the images. The adversarial suppression branch, embedded with two occlusion suppression module, minimizes the generated occlusion's response and strengthens attentive feature representation on human non-occluded body regions. Finally, we get a more discriminative pedestrian feature descriptor by concatenating two branches' features, which is robust to the occlusion problem. The experiments on the occluded dataset show the competitive performance of IGOAS. On Occluded-DukeMTMC, it achieves 60.1% Rank-1 accuracy and 49.4% mAP.

Movement fit design aspects of bicycle shorts

The variables influencing the functional requirements of bicycle shorts in terms of providing support for movement were studied. In order to identify the movement fit design characteristics that specify the state of mechanical movement comfort for bicycle shorts used in indoor fit regimens and weight loss training, the pattern of interaction between the clothing and human body determined by 2D pattern proportions and fabric 's stretch property is examined. The study was performed in the scope of bio-mechanical movement comfort, a branch of ergonomics by observing stretch deformation and shear deformation during wear. In the context of the ergonomic view point, the study was carried out by verifying the compression pressure imparted by the clothing and analyzing the nature of stretch deformation across waist hip regions of human body. The role of fabric stretch property and concept of 2D pattern suppression that facilitates ideal mechanical fit design characteristics was ascertained through the analysis. Further the concept of pattern suppression’s influence in achieving more uniform compression pressure distribution is also concluded.

Discrete pointvv 3D reconstruction algorithm based human pose estimation

The traditional method of human pose estimation lacks depth label in depth image, which leads to the low generalization ability of model due to single pose. A novel method of human pose estimation based on discrete point 3D reconstruction algorithm is proposed. Several cameras are placed around each other, and the internal parameters of the camera are calibrated by the classical calibration method , and the external parameters of the camera group are obtained. According to the two-dimensional skeleton, the human body region model is established, and the contour is matched with the human body model, so that the contour region is segmented, and then the three-dimensional human body model is constructed. According to the principle of discrete point 3D reconstruction algorithm, principal component analysis is used to transform the original data, eliminate the redundant correlation between the original human body posture data, establish the human body posture model, obtain 10 individual voxel sets, and estimate the 3D information of each point in the voxel set. In the process of feature extraction, two-dimensional attitude prediction, two-dimensional fusion layer and final conversion, the human body attitude estimation is realized. Experimental results show that this method can reduce the influence of image shadow on image recognition, and the estimation results are more accurate.

TML: A Triple-Wise Multi-Task Learning Framework for Distracted Driver Recognition

We propose a multi-task learning framework for improving the performance of vision-based deep-learning approaches for driver distraction recognition. The most popular tool so far for solving this task is convolutional neural networks (CNNs) that have proven to be strongly biased toward local features. Such bias causes CNNs to neglect global structural information, adversely affecting the robustness of the distracted driver recognition task. To solve this problem, we generate positive and negative samples of each given input, and construct a triplet of images (i.e., raw image, positive sample, and negative sample). The positive sample is generated by applying structure-aware illumination to the human body region of each given input. The negative sample is generated by randomly shuffling the local regions of each given input. The networks are then trained with the triplets using a multi-task learning strategy to force the networks to explore global information by multiple tasks: (a) recognizing the raw input and positive sample as the given ground truth; (b) recognizing the negative sample as an extra “meaningless” label; (c) pulling closer the distance between the features obtained from the raw input and positive sample while pushing away the distance between the features obtained from the raw input and negative sample. By doing so, the model can be trained so that it neglects the background information and pays more attention to the global structual information of the scene. The proposed approach reaches state-of-the-art performance on the AUC Distracted Driver Dataset and performs better than state-of-the-art studies on the Drive and Act Dataset. With raw images as input, we have achieved an accuracy of 96.0% for the AUC distracted driver dataset and 66.8% for the Drive and Act Dataset. Our approach does not introduce extra overhead during the testing procedure (i.e., utilization procedure), which is helpful for real-life applications. Moreover, better accuracy can be achieved by fusing the predictions respectively obtained with the raw input and positive sample. As a result, we have achieved an accuracy of 96.3% for the AUC distracted driver Dataset and 66.9% for the Drive and Act Dataset. The class activation map (CAM) of our proposed method is subjectively more reasonable, which would enhance the reliability and explainability of the model.

Mask‐guided class activation mapping network for person re‐identification

In this Letter, the authors propose a novel mask-guided class activation mapping (MCAM) network for person re-identification, which learns background-invariant and view-invariant features. Specifically, a novel loss function named mask-guided mapping loss is meticulously formulated to utilise the human binary masks, which contain helpful body shape information as the reference standard, thereby guiding the model to place more emphasis on human body regions. Moreover, they propose a new weighted channel attention (WCA) module, which replaces the global average pooling with a global depthwise convolution layer. By virtue of this particular WCA module, the feature information distributed across the spatial space can be individually weighted and dynamically compressed into a more precise channel attention map. Extensive experiments have been carried out on three widely-used re-identification data sets. Compared with the baseline model, MCAM has gained rank-1 accuracy improvement of 2.0% on Market-1501, 6.0% on DukeMTMC-reID, and 7.5% on CUHK03-NP, confirming its effectiveness.