- Research Article
- 10.1007/s12204-026-2898-9
Advancing Speech Data Collection and Annotation: General Methods, Practical Experience, and Future Perspectives
- Jan 28, 2026
- Journal of Shanghai Jiaotong University (Science)
- Yingming Gao + 5 more +5
Publications from 2021 to 2026
Showing 10 of 55 papers
Advancing Speech Data Collection and Annotation: General Methods, Practical Experience, and Future Perspectives
Photoredox-catalysed cyclisation of carbyne equivalents with alkenes and aldehydes to dihydrofurans.
Herein, we report a multicomponent cyclisation reaction that employs a diazomethyl-substituted hypervalent iodine reagent as a carbyne source in combination with simple alkenes and aldehydes. This strategy successfully constructs dihydrofuran scaffolds, featuring broad substrate scope and excellent functional group compatibility. This work underscores the utility of carbyne as a versatile intermediate in organic synthesis.
Read moreResearch on plug-and-play scheme of relay protection device in substation
Relay protection devices are key equipment to ensure the safe and stable operation of power systems. The current design schemes of device and panel have some problems such as long power outages, heavy workload, and potential safety hazard of wiring during the transformation process, which are in sharp conflict with the requirements of safe power supply and high reliability of the power grid. Several conventional transformation schemes of the panel are presented in this paper. And a technical solution based on a new type of terminal block is proposed, aiming to achieve universal interchangeability and plug-and-play of full-voltage level protection devices, enhance power supply reliability and improve operation and maintenance convenience. It provides a method for the rapid transformation of relay protection devices.
Read moreDynamic performance of composite laminate wing beam structures of aeronautical engineering with complex geometrical profiles
Design and analysis of a simultaneous wireless power and data transfer system across insulators for online monitoring devices on transmission lines
To enhance the operational reliability of wireless power supply systems for transmission line monitoring devices and to facilitate bidirectional communication between the power transmitter and receiver, this study proposes a simultaneous wireless power and data transfer (SWPDT) scheme employing a multi-relay coil configuration. By exploiting the multi-resonant behavior inherent in high-order compensation topologies—specifically LCC and LCL structures—the proposed system realizes efficient co-transmission of energy and information. A comprehensive power transmission model, alongside an equivalent circuit model for data communication, is developed to capture the electromagnetic coupling and transmission dynamics within the multi-relay environment. The characteristics of both power and data channels are analytically examined, and the validity of the proposed methodology is substantiated through simulation studies. The results confirm that the system achieves robust power delivery while maintaining reliable data transmission, thereby offering a promising solution for deployment in high-voltage insulator string scenarios.
Read moreA Flexible Load Precision Control Device’s Autonomous Method for Station Areas Based on Edge-Computing
With the advancement of new power system construction, the large-scale integration of renewable energy has significantly increased the demand for distributed regulation. This paper proposes a flexible load precision control device’s autonomous method for station areas based on edge-computing. The primary objectives are to control the maximum load rate of the distribution area, reduce reverse power flow from the area to the grid, and provide reserve capacity for the grid. An optimized operation model for autonomous management is developed to generate an optimal operation plan for the distribution area. Additionally, a real-time control model is established to dynamically adjust the operation when the planned operation results in heavy loads or reverse power flow to the grid. Both models, along with a CBC solver for optimization, are deployed within the flexible load control device. This method mitigates the impact of errors in load and renewable energy power forecasts, enhances the safety of both the distribution area and the grid, and reduces the need for capacity upgrades or replacements of station areas.
Read moreEdge-Guided Human Depth Image Inpainting Based on Diffusion Models
In the context of depth images, fusing human segmentation images with depth background images provides critical data for human detection tasks, where accurate localization of the human head is a key indicator. However, in real-world scenarios with occlusions and complex backgrounds, human segmentation images often lack head pixels. This missing data leads to subpar performance in human recognition tasks for deep learning models. To address this issue, this paper proposes an efficient approach that combines edge detection, projection analysis, and template matching for upper-body localization and head identification. Additionally, a refined diffusion model is used to repair the human head edge mask obtained from the initial processing. The main advantage of this approach is its independence from complex deep learning model training. It integrates traditional image processing techniques, morphological analysis, and diffusion models, simplifying head localization and effectively avoiding interference from the shoulder and arm regions. This approach enables more accurate head localization and uses diffusion models to restore missing portions of the human head, thereby improving overall pose recognition accuracy. Experimental results demonstrate that the repaired depth images outperform the original images in human pose model training. When trained with the repaired images, detection accuracy (mAPVal50-95) and precision improved by 10.1 % and 7.1%, respectively, significantly enhancing downstream object detection performance. These findings validate the effectiveness of the proposed method in repairing missing human head depth images. They further highlight its potential in improving pose recognition and object detection performance, offering reliable technical support for related tasks.
Read moreMultimodal Sign Language and Big Data Integration Research
Sign language is the primary means of communication for the deaf and mute community. The regional and national differences in sign language have led to barriers between different sign language systems, making communication between deaf and mute individuals from different regions challenging. This paper proposes a vision-tactile fusion approach to recognize sign language. The scheme applies a flexible sensor data glove to monitor the angle changes of finger joints, as well as visual image processing technology, to integrate tactile and visual data for sign language recognition. The integration is performed through three methods: weighted average, weighted multiplication, and attention weights, with accuracy rates of 95.55%, 97.87%, and 98.83%, respectively.
Read moreResearch on Robot Path Planning and Localization Based on ROS
Recently, the rapid development of robotics technology has made enhancing autonomous navigation capabilities in complex environments increasingly essential. Path planning is key to improving the autonomy of robots, while localization is the process of determining the robot’s accurate position within the environment. Given this, this research delves into path planning and localization technologies based on the Robot Operating System (ROS), comparing and analyzing path planning algorithms such as Dijkstra, A*, RRT and its variants, as well as particle filter localization techniques. By discussing the performance of these technologies in practical applications, we highlight their strengths and weaknesses in high-dimensional spaces and complex terrains and proposes corresponding optimization strategies to enhance robots’ navigation capabilities in complex environments. The research indicates that the appropriate combination of various path planning and localization technologies in different application scenarios can effectively improve the performance of robotic systems, providing a theoretical basis and practical guidance for the optimization of robotic navigation systems.
Read moreDesign of labor capacity management system based on artificial intelligence and machine learning
With the development of artificial intelligence technology and the expansion of its application scope, an intelligent change has taken place in human resource management. Due to the influence of human factors and experience, the traditional manpower allocation method is difficult to achieve the best benefits. On the basis of artificial intelligence, using machine learning, big data analysis and other methods, human resources can be optimized to the maximum extent, and the operating efficiency and production efficiency of enterprises can be improved. This paper makes an in-depth analysis of the imbalance between supply and demand of human resources and the low efficiency of resource allocation in human resources management in China. Through the collection and analysis of massive human resources, this system uses machine learning method to identify and predict them, which can accurately judge the employment demand of each position, so as to achieve the purpose of automatic deployment. In this paper, the main functional modules, data processing flow and key technologies of the system are discussed in detail. Combined with specific examples, this paper makes an empirical study on the adopted methods. A company (manufacturing) reduced the labor cost by 800,000 yuan, while C company (manufacturing) reduced the labor cost by 1.5 million yuan. The research shows that using artificial intelligence technology to allocate human resources can effectively improve the efficiency and accuracy of human resources allocation, reduce human disturbance and reduce the operating cost of enterprises.
Read more