Computer Animation and Virtual Worlds最新文献

Labanotation is a scientific method for documenting dance movements that has been widely adopted globally. Existing methods for Labanotation action recognition perform poorly in handling complex movements and integrating spatiotemporal information. To address this, we propose a multi-branch spatiotemporal fusion network with attention mechanisms aimed at accurately recognizing Labanotation actions from motion capture data. Initially, we convert motion capture data into three-dimensional coordinates and extract skeleton vector features. Subsequently, we enhance feature representation by extracting temporal difference features and skeleton angle features from the skeleton vectors. These features are processed using gated recurrent units and residual networks to effectively integrate spatiotemporal information. Finally, attention mechanisms are applied in the model to differentiate the importance of different positions in the features. This method effectively models spatiotemporal relationships, thereby improving the accuracy of Labanotation action recognition. We conducted experiments on two segmented motion capture datasets, demonstrating the effectiveness of each module. Compared to existing methods, our approach shows superior performance and strong generalization ability. Given the relative simplicity of upper limb action recognition, our focus primarily lies on lower limb action recognition. Notably, this marks the first application of skeleton angle features in the field of Labanotation action recognition.

Though current traffic simulation methods can produce impressive results, reversing behavior is always ignored, potentially reducing the diversity and plausibility of simulation data. Furthermore, while common traffic behaviors like following-the-leader and lane changing can be easily simulated, efficiently generating irregular cases in a human-in-the-loop manner with specific motions based on user desires is less discussed. To address the gap, we present a novel interactive traffic editing and simulation framework that enables users to regulate vehicles via simple inputs to introduce reversing and generate desired trajectory data with both car-following and irregular driving behaviors. With key states specified, lane-level navigation, including forward/backward directions, is planned through heuristic search. The customized navigation brings the vehicles' new trajectories with both car-following and reversing, and their surrounding neighbors are also adjusted accordingly. To provide smooth and plausible motions after editing, vehicles are updated via the optimization-based simulation method, which takes vehicle kinematics, self-motivation, path keeping, collision avoidance, and special interaction rules into account. We demonstrate that our framework can generate uncommon traffic cases and validate it through extensive experiments.

Procedural modeling methods are used to automatically generate virtual scenes. There is a large number of available top-down methods for generating partial content for specific purposes. However, little research was done on enabling the generation of content in the presence of manually modeled elements, from the bottom-up direction, or without significant assistance from the user. No existing approach provides a platform that can combine the results of different methods, which leaves them isolated. This paper presents an integration approach that generates complete virtual space organizations by combining the usage of top-down and bottom-up procedural generation of content, with support for the placement of manually modeled content. The integration is made possible by using shape conversion to match the input and output shape types of different methods. The evaluation of the proposed approach was performed on a 2D polygon dataset by using four different scenarios, validating that it works as intended. Additional testing was performed by using a case study of organizing 3D virtual space around the manually modeled element of virtual heritage Tašlihan to demonstrate all capabilities of the integration approach and the different outputs depending on the level of user interaction and the desired results.

This study presents an innovative approach to dragline mine operations using 3D virtual reality technology, specifically developed at the virtual reality mines simulation facility at IIT (ISM), Dhanbad. Focused on Northern Coalfields Limited, Singrauli, this research opens up the use of immersive VR for virtual inspection and analysis of dragline operations, aiding its effective deployment. The methodology involves capturing geo-spatial data using drones and integrating it into a commissioned dragline simulator workbench. The integration process combines drone imagery to accurately reconstruct the physical mine site. Enhanced by the Structure-from-Motion technique, the resulting model is a photorealistic point cloud, offering unprecedented visualization accuracy. The developed system enables complex manipulation of datasets, including functions such as scaling, rotation, and translation. It also includes geometric measurement tools for determining length, area, and volume, essential for precise operational planning of draglines. The application of Structure-from-Motion-MultiView Stereo technology is particularly noteworthy for its role in tracking safety concerns and monitoring dragline progress with respect to the stipulated balance diagram. The proposed approach surpasses traditional mine visualization methods, providing superior tools for onsite planning and comprehensive asset management and significant contributions for establishing a new benchmark for monitoring and visualizing the dragline excavation process in the mining industry.

The current 3D modeling methods for analysis, documentation, and preservation of cultural heritage sites are tedious, and the realistic expression ability of the model is insufficient. It is urgent to break through the bottleneck of professional model creation and architectural layout analysis by non-professional users. Based on the study of architectural regulations of Confucian temples, this paper presents an HBIM (Heritage Building Information Modeling) framework for management and spatial analysis of the historical buildings within the space of Confucian temples. This paper solves the problems of low degree of automation in 3D modeling and lack of detail expression ability in virtual architectural layout analysis for Confucian temples. The framework can be an important tool to analyze the architectural space form. Through case studies of Confucian temples in Qufu and Beijing, the impact of the study on enhancing non-professional users' modeling experience is revealed. The results of the study strengthen the digital presentation of structure and hierarchy of ancient Chinese society, thus become the starting point of further research for Confucian temples all over the world.

The underground coal mine working environment is complicated, and the detection of safety helmet wearing is vital for assuring worker safety. This article proposes an improved YOLOv8n safety helmet detection model, YOLOv8-HAC, to address the issues of coexisting strong light exposure and low illumination, equipment occlusions that result in partial target loss, and the missed detection of small targets due to limited surveillance perspectives in underground coal mines. The model substitutes the suggested HAC-Net for the C2f module in YOLOv8n's backbone network to improve feature extraction and detection performance for targets with motion blur and low-resolution images. To improve detection stability in complicated situations and lessen background interference, the AGC-Block module is also included for dynamic feature selection. Additionally, a tiny target detection layer is included to increase the long-range identification rate of tiny safety helmets. According to experimental data, the enhanced model outperforms existing popular object detection algorithms, with a mAP of 94.8% and a recall rate of 90.4%. This demonstrates how well the suggested approach works to identify safety helmets in situations with complicated lighting and low-resolution photos.

This paper introduces a cloud-edge collaborative framework for real-time virtual-real fusion rendering in augmented reality (AR). By integrating Visual Simultaneous Localization and Mapping (VSLAM) with Neural Radiance Fields (NeRF), the proposed method achieves high-fidelity virtual object placement and shadow synthesis in real-world scenes. The cloud server handles computationally intensive tasks, including offline NeRF-based 3D reconstruction and online illumination estimation, while edge devices perform real-time data acquisition, SLAM-based plane detection, and rendering. To enhance realism, the system employs an improved soft shadow generation technique that dynamically adjusts shadow parameters based on light source information. Experimental results across diverse indoor environments demonstrate the system's effectiveness, with consistent real-time performance, accurate illumination estimation, and high-quality shadow rendering. The proposed method reduces the computational burden on edge devices, enabling immersive AR experiences on resource-constrained hardware, such as mobile and wearable devices.

In this article, we propose an AI-agent that integrates a large language model (LLM) with a retrieval-augmented generation (RAG) system to deliver reliable historical information from the Annals of the Joseon Dynasty through both objective facts and contextual analysis, achieving significant performance improvements over existing models. For an AI-agent using the Annals of the Joseon Dynasty to deliver reliable historical information, clear source citations and systematic analysis are essential. The Annals, an official record spanning 472 years (1392–1897), offer a dense, chronological account of daily events and state administration that shaped Korea's cultural, political, and social foundations. We propose integrating a LLM with a RAG system to generate highly accurate responses based on this extensive dataset. This approach provides both objective information about historical figures and events from specific periods and subjective contextual analysis of the era, helping users gain a broader understanding. Our experiments demonstrate improvements of approximately 23 to 50 points on a 100-point scale compared with the GPT-4o and OpenAI AI-Assistant v2 models.