Front View Gait Recognition using Spherical Space Model with Human Point Clouds

In this paper, we propose a novel gait recognition framework - Spherical Space Model with Human Point Clouds (SSM-HPC) to recognize front view of human gait. A new gait representation - Marching in Place (MIP) gait is also introduced which preserves the spatiotemporal characteristics of individual gait manner. In comparison with the previous studies on gait recognition which usually use human silhouette images from image sequences, this research applies three dimensional (3D) point clouds data of human body obtained from stereo camera. The proposed framework exhibits gait recognition rates superior to those of other gait recognition methods.

In complex industrial human-robot collaboration (HRC) environment, obstacles in the shared working space will occlude the operator, and the industrial robot will threaten the safety of the operator if it is unable to get the complete human spatial point cloud. This paper proposes a real-time human point cloud inpainting method based on the deep generative model. The method can recover the human point cloud occluded by obstacles in the shared working space to ensure the safety of the operator. The method proposed in this paper can be mainly divided into three parts: (i) real-time obstacles detection. This process can detect obstacle locations in real time and generate the image of obstacles. (ii) the application of the deep generative model algorithm. It is a complete convolutional neural network (CNN) structure and introduces advanced generative adversarial loss. The model can generate the missing depth data of operators at arbitrary position in the human depth image. (iii) spatial mapping of the depth image. The depth image will be mapped to point cloud by coordinate system conversion. The effectiveness of the method is verified by filling hole of the human point cloud occluded by obstacles in industrial HRC environment. The experiment results show that the proposed method can accurately generate the occluded human point cloud in real time and ensure the safety of the operator.

Current point cloud extraction methods based on photogrammetry generate large amounts of spurious detections that hamper useful 3D mesh reconstructions or, even worse, the possibility of adequate measurements. Moreover, noise removal methods for point clouds are complex, slow and incapable to cope with semantic noise. In this work, we present body2vec, a model-based body segmentation tool that uses a specifically trained Neural Network architecture. Body2vec is capable to perform human body point cloud reconstruction from videos taken on hand-held devices (smartphones or tablets), achieving high quality anthropometric measurements. The main contribution of the proposed workflow is to perform a background removal step, thus avoiding the spurious points generation that is usual in photogrammetric reconstruction. A group of 60 persons were taped with a smartphone, and the corresponding point clouds were obtained automatically with standard photogrammetric methods. We used as a 3D silver standard the clean meshes obtained at the same time with LiDAR sensors post-processed and noise-filtered by expert anthropological biologists. Finally, we used as gold standard anthropometric measurements of the waist and hip of the same people, taken by expert anthropometrists. Applying our method to the raw videos significantly enhanced the quality of the results of the point cloud as compared with the LiDAR-based mesh, and of the anthropometric measurements as compared with the actual hip and waist perimeter measured by the anthropometrists. In both contexts, the resulting quality of body2vec is equivalent to the LiDAR reconstruction.

Surveillance systems have long been in the focus of the research community. Although the accurate detection of the human presence in the scene is now possible even under extreme environmental conditions via the advanced modern camera sensors, efficient personalized tracking is still an open issue and a significant challenge for researchers addressing. Moreover, personalized tracking will not only enhance the tracking robustness but it can also find useful application in several commercial surveillance use-cases, ranging from security to occupancy statistics (i.e. per building, per space and per human). In this respect, this paper introduces a novel the biometric approach for enhanced privacy preserving human tracking based on a novel soft-biometric feature of humans. The moving blobs in the recorded scene can be easily detected in the colour images, while the human silhouettes are detected from the corresponding depth ones. The state-of-the-art 3D Weighted Walkthroughs (3DWW) transformation is applied on the extracted human 3D point cloud, forming thus, a short-term soft biometric signature. The re-authentication of the humans is performed via the comparison of their last valid signature with current one. A thorough analysis on the adjustment of the system’s optimal operational settings has been carried out and the experimental results illustrate the promising robustness, accuracy and efficiency on human tracking performance.

Combining single-pixel imaging with frequency modulated continuous wave radar in the submillimeter wavelength regime provides a promising alternative to high cost detector arrays for 3D surface imaging. We investigate how compressive sensing reconstructions can be applied for this specific scenario. An iterative reconstruction method is proposed and numerical experiments with simulated data based on a human 3D point cloud are performed. Further, we present a model extension for exploiting additional knowledge about the depth information during recovery.

Optimization of gait features using Locally Linear Embedded with Multi-layer Perceptron for frontal view is explored in this study. Static gait features within a gait cycle are extracted from gait data extracted using Kinect. The extracted features are further optimized using Locally Linear Embedded and classified using Multi-layer Perceptron. To verify the effectiveness of the proposed method, original features are also utilised. Result showed that the recognition of human gait using Multi-layer Perceptron with 30 hidden units along with optimal feature of Locally Linear Embedded with K = 88 and d = 64 outshined the recognition rate specifically 98%.

Solar wind forecasting plays a crucial role in space weather prediction, yet significant uncertainties persist duet to incomplete magnetic field observations of the Sun. Isolating the solar wind forecasting errors due to these effects is difficult. This study investigates the uncertainties in solar wind models arising from these limitations. We simulate magnetic field maps with known uncertainties, including far-side and polar field variations, as well as resolution and sensitivity limitations. These maps serve as input for three solar wind models: the Wang–Sheeley–Arge, the Heliospheric Upwind eXtrapolation, and the European Heliospheric FORecasting Information Asset. We analyze the discrepancies in solar wind forecasts, particularly the solar wind speed at Earth’s location, by comparing the results of these models to a created ground truth magnetic field map, which is derived from a synthetic solar rotation evolution using the Advective Flux Transport model. The results reveal significant variations within each model with a root mean square error ranging from 59 to 121 km s−1. Further comparison with the thermodynamic Magnetohydrodynamic Algorithm outside a Sphere model indicates that uncertainties in the different models can lead to even larger variations in solar wind forecasts compared to those within a single model. However, predicting a range of solar wind velocities based on a cloud of points around Earth can help mitigate uncertainties by up to 20%–77%.

Chapter 1 - Projective geometry

Detailed forest inventory and mensuration of individual trees have drawn attention of research society mainly to support sustainable forest management. This study aims at estimating individual tree attributes from high density point cloud obtained by terrestrial laser scanner (TLS). The point clouds were obtained over single reference tree and group of trees in forest area. The reference tree is treated as benchmark since detailed measurements of branch diameter were made on selected branches with different sizes and locations. Diameter at breast height (DBH) was measured for trees in forest. Furthermore tree height, height to crown base, crown volume and tree branch volume were also estimated for each tree. Branch diameter is estimated directly from the point clouds based on semi-automatic approach of model fitting i.e. sphere, ellipse and cylinder. Tree branch volume is estimated based on the volume of the fitted models. Tree height and height to crown base are computed using histogram analysis of the point clouds elevation. Tree crown volume is estimated by fitting a convex-hull on the tree crown. The results show that the Root Mean Squared Error (RMSE) of the estimated tree branch diameter does not have a specific trend with branch sizes and number of points used for fitting process. This explains complicated distribution of point clouds over the branches. Overall cylinder model produces good results with most branch sizes and number of point clouds for fitting. The cylinder fitting approach shows significantly better estimation results compared to sphere and ellipse fitting models.

Gait recognition, crucial in biometrics and behavioral analytics, has applications in human–computer interaction, identity verification, and health monitoring. Traditional sensors face limitations in complex or poorly lit settings. RF-based approaches, particularly millimeter-wave technology, are gaining traction for their privacy, insensitivity to light conditions, and high resolution in wireless sensing applications. In this paper, we propose a gait recognition system called Multidimensional Point Cloud Gait Recognition (PGGait). The system uses commercial millimeter-wave radar to extract high-quality point clouds through a specially designed preprocessing pipeline. This is followed by spatial clustering algorithms to separate users and perform target tracking. Simultaneously, we enhance the original point cloud data by increasing velocity and signal-to-noise ratio, forming the input of multidimensional point clouds. Finally, the system inputs the point cloud data into a neural network to extract spatial and temporal features for user identification. We implemented the PGGait system using a commercially available 77 GHz millimeter-wave radar and conducted comprehensive testing to validate its performance. Experimental results demonstrate that PGGait achieves up to 96.75% accuracy in recognizing single-user radial paths and exceeds 94.30% recognition accuracy in the two-person case. This research provides an efficient and feasible solution for user gait recognition with various applications.

We characterize the topology of the phase space of the Berlin-Kac spherical model in the context of the so called Topological Hypothesis, for spins lying in hypercubic lattices of dimension d. For zero external field we are able to characterize the topology exactly, up to homology. We find that, even though there is a continuum of changes in the topology of the corresponding manifolds, for d ≥ 3 there are abrupt discontinuities in some topological functions that could be good candidates to associate with the phase transitions that occur at the thermodynamic level. We show however that these changes do not coincide with the phase transitions and conversely, that no topological discontinuity can be associated to the points where the phase transitions take place. At variance with what happens in the Mean Field version of this same model, we show that these abrupt topological changes are accessible thermodynamically. We conclude that, even in short range systems, the topological mechanism does not seem to be responsible for the triggering of a phase transition. We also analyze the case of spins connected to a macroscopic number of (but not all) neighbors, and find that, similar to the results found for the fully connected version, in this case the topological hypothesis seems to hold: the phase transition coincides with an accumulation point of the topological changes present in configuration space. The question of the ensemble equivalence in the short range spherical model is also considered.

In the Lobachevsky hyperbolic and Riemann spherical spaces, generalized potentials describing a uniform electric field are introduced as solutions of the covariant Maxwell equations. Exact solutions of the Schr¨odinger equation in the presence of the electric field are constructed in both models. The similarity of the energy spectra of the particle against the background of a spherical space with the electric field and in the Coulomb field is noted.

Large-scale terrestrial laser scanning (TLS) point clouds are increasingly used for applications such as digital twins and cultural heritage documentation; however, removing unwanted human points captured during acquisition remains a largely manual and time-consuming process. This study proposes a geometry-aware framework for automatically removing human noise from TLS point clouds by projecting 2D instance segmentation masks (obtained using You Only Look Once (YOLO) v8 with an instance segmentation head) into 3D space and validating candidates through multi-stage geometric filtering. To suppress false positives induced by reprojection misalignment and planar background structures (e.g., walls and ground), we introduce projection-followed geometric validation (or "geometric gating") using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and principal component analysis (PCA)-based planarity analysis, followed by cluster-level plausibility checks. Experiments were conducted on two real-world outdoor TLS datasets-(i) Osaka Metropolitan University Sugimoto Campus (OMU) (82 scenes) and (ii) Jinaimachi historic district in Tondabayashi (JM) (68 scenes). The results demonstrate that the proposed method achieves high noise removal accuracy, obtaining precision/recall/intersection over union (IoU) of 0.9502/0.9014/0.8607 on OMU and 0.8912/0.9028/0.8132 on JM. Additional experiments on mobile mapping system (MMS) data from the Waymo Open Dataset demonstrate stable performance without parameter recalibration. Furthermore, quantitative and qualitative comparisons with representative time-series geometric dynamic object removal methods, including DUFOMap and BeautyMap, show that the proposed approach maintains competitive recall under a human-only ground-truth definition while reducing over-removal of static structures in TLS scenes, particularly when humans are observed in only one or a few scans due to limited revisit frequency. The end-to-end processing time with YOLOv8 was 935.62 s for 82 scenes (11.4 s/scene) on OMU and 571.58 s for 68 scenes (8.4 s/scene) on JM, supporting practical efficiency on high-resolution TLS imagery. Ablation studies further clarify the role of each stage and indicate stable performance under the observed reprojection errors. The annotated human point cloud dataset used in this study has been publicly released to facilitate reproducibility and further research on human noise removal in large-scale TLS scenes.

In this paper, we propose a key pose based gait recognition approach using skeleton joint information derived from the depth data of Kinect. We consider situations where such depth cameras are mounted on top of entry and exit points, respectively capturing back and front views of subjects who enter a zone under surveillance. Three dimensional geometric transformations are used to map the skeleton images captured from the back view to an equivalent front view. A gait cycle is divided into a number of key poses and the trajectory followed by each skeleton joint within a key pose is used to derive the gait features for that particular pose. For recognizing a subject, available key poses are compared with the corresponding key poses of the training subjects. The proposed method has higher accuracy than other competing approaches.

Conventional text embeddings, typically learned in the Euclidean space, may struggle to effectively capture word semantics based solely on the directional similarity between word vectors. In contrast, spherical text embeddings have demonstrated remarkable efficacy in various natural language processing (NLP) tasks recently. However, training spherical text generative models (STGMs) requires large representative datasets, which could potentially contain sensitive private information. To mitigate this concern, we propose a novel approach: the differential private spherical text generative model (DP-STGM), which facilitates learning text embeddings within the spherical space while ensuring privacy via efficient Riemannian optimization and the framework of differential privacy. To evaluate the efficacy of our privacy-preserving algorithm, we initially train an adversary using an external dataset without the application of differential privacy. Subsequently, we introduce two metrics to measure the model’s ability to protect privacy: (1) cosine similarity between recovered words from the adversary that generates the spherical text embeddings and those generated by DP-STGM, and (2) Top-n rank correlation. Our experimental findings demonstrate that DP-STGM outperforms baseline models, showcasing its superior performance. By leveraging the power of differential privacy in the Riemannian optimization process, our model achieves better preservation of sensitive information while simultaneously capturing the semantic nuances inherent in word embeddings. As a result, DP-STGM represents a robust and efficient solution for NLP tasks that require privacy protection without compromising on the quality of learned text embeddings. By offering a privacy-preserving alternative, DP-STGM broadens the range of applications for which STGMs can be safely employed, ensuring data privacy while harnessing the rich information contained in spherical text embeddings. Our work opens up new avenues for future research in privacy-aware NLP and advances the state-of-the-art in both privacy protection and semantic learning in this domain.