Given the critical role of birds in ecosystems, Fine-Grained Bird Recognition (FGBR) has gained increasing attention, particularly distinguishing within similar subcategories. Although Vision Transformer (ViT)-based methods often outperform Convolutional Neural Network (CNN)-based FGBR, recent studies reveal that limited receptive field plain ViT model hinders representational richness and makes them vulnerable to scale variance. Thus, enhancing multi-scale capabilities existing ViT-based...

In order to understand how geometric errors propagate and deviations accumulate in a six-axis motion platform (SMP), new error model based on the stream of variation (SOV) theory is presented this paper. SOV widely used industrial engineering with several steps or phases. The conventional only calculates initial final orientation, yet after each step whole process are still unknown, which critical parameters measuring product quality. modeling method, alignment welding SMP can be considered...

To improve high motion accuracy and efficiency in the high-speed operation of a 4-DOF (4 degrees freedom) redundant parallel robot, this paper introduces trajectory planning robot joint space based on twelve-phase sine jerk profile. The 12-phase profile utilizes characteristics function. Furthermore, penalty function is used to optimize energy consumption under constraint condition. simulation experimental results show that slightly higher than three-phase profile, but overall more accurate...

Human motion generation is a cut-edge area of research in generative computer vision, with promising applications video creation, game development, and robotic manipulation. The recent Mamba architecture shows results efficiently modeling long complex sequences, yet two significant challenges remain: Firstly, directly applying to extended ineffective, as the limited capacity implicit memory leads decay. Secondly, struggles multimodal fusion compared Transformers, lack alignment textual...

Aiming at problems of planar optical waveguide aligning and coupling multi-dimensional space sub-micron motion error, with using topology theory, estimating distribution error paramount precisely, this paper establishes position model precise platform, based on Monte Carlo method, obtains fiber end point. The result indicated that method can obtain accurate in the adjust process automatic coupling, achieve effective control accuracy.

This paper introduces a new analyzing approach on positioning accuracy (PA) in precise linear motion stage (PLMS).As one of the most important parameter, PA is affected by several reasons for quality PLMS, such as machining error, assembly error and controller etc. Conventional method only focuses aspect leading to (PE), which not systematically comprehensively.By adopting approach, it obviously understandable how errors from each part PLMS give impact PE accumulates.In this paper, measuring...

Background: Early identification of postoperative ischemic stroke (PIS) for acute DeBakey type I aortic dissection (ADIAD) patients is great significance to taking effective treatment in time. We aim develop and validate a prediction model PIS ADIAD who underwent the surgery total arch replacement (TAR) frozen elephant trunk (FET) under mild hypothermia.Methods: From January 2017 April 2023, we enrolled TAR FET our study. Preoperative intraoperative variables were selected using pairwise...

<p>To improve high motion accuracy and efficiency in the high-speed operation of a 4-DOF (4 degrees freedom) redundant parallel robot, this paper introduces trajectory planning robot joint space based on twelve-phase sine jerk profile. The 12-phase profile utilizes characteristics function. Furthermore, penalty function is used to optimize energy consumption under constraint condition. simulation experimental results show that slightly higher than three-phase profile, but overall more...

<p>To improve high motion accuracy and efficiency in the high-speed operation of a 4-DOF (4 degrees freedom) redundant parallel robot, this paper introduces trajectory planning robot joint space based on twelve-phase sine jerk profile. The 12-phase profile utilizes characteristics function. Furthermore, penalty function is used to optimize energy consumption under constraint condition. simulation experimental results show that slightly higher than three-phase profile, but overall more...

CNC machining center is an indispensable equipment for precision machining. The current of still cannot meet the requirements, depth error tracing not enough, and workpiece needs to be improved. Therefore, this paper proposes a method based on distribution feature recognition. Taking three-axis as example, carried out. probability density function relationship between "component/part - error" derived through "stream variation" theory. grooves length parameter was measured by ultra-depth...

The multi-axis motion system(MMS) includes optoelectronic packaging system and CNC machine tools, the quality of which mainly rely on precision its controlling system. distribution influential ability errors in this needs high consideration to address improvement problem. This issue has become more complex due their strict high-accuracy requirements. It is therefore increasingly important characterize interaction between configuration setting MMS. In paper, methodology analytical procedures...