Large Objects Research Articles

X-CT Reconstruction as a Tool for Monitoring the Conservation State and Decay Processes of Works of Art and in Support of Restoration and Conservation Strategies

X-ray Computed Tomography (X-CT) is now an established technique for the investigation and diagnostics of Cultural Heritage. Its advantages include non-invasiveness, non-destructiveness, and the possibility of exploring the inner parts of an object without any modification. X-CT is often employed to investigate the construction methods of complex artifacts made with different parts or materials, but it is also able to support the analysis, intervention, monitoring and enhancement processes of artworks, creating digital models that can aid in the conservation and restoration procedures. In this work, several case studies are presented in which the CT technique has been decisive in identifying the effects of time and the events that occurred during the object’s life influencing its state of conservation. These range from large objects, such as an 18th century CE writing cabinet or an ancient Egyptian wooden coffin, to very small artifacts, like Mesopotamian lapis lazuli beads or fragments of Roman colored glass. Additionally, the results obtained by µ-CT investigations on the conservation state of a bronze arrowhead uncovered from the Urama-chausuyama mounded tomb (Japan, Kofun period, end of the 3rd century CE) are presented here for the first time. Lastly, the versatility of the technique when applied with different setups is highlighted.

Ceres: Organic‐Rich Sites of Exogenic Origin?

AbstractCeres, the largest object in the asteroid belt, is the only potential ocean world in the inner Solar System. Previous studies identified deposits of aliphatic organics in and around the Ernutet crater, and at small locations at Inamahari and Urvara craters. The origin of organics, either endogenic or exogenic, in these fresh exposures is still under debate. This study addresses the origin of the organics by analyzing their global distribution and geologic context. Our first step involved a global search for organic‐rich sites that might have escaped previous detections. We achieved this by using a deep neural network, utilizing spectral redness in the Dawn's Framing Camera multispectral data to identify potential organic‐rich sites. The identified sites were further studied by using IR spectrometer data to infer the compositions of materials showing spectral redness. Of the newly identified red‐sloped sites, only two can be considered certain to be organic‐rich. We also identified sites with spectral redness, but without any signature of organics in their infrared spectra. These sites could be attributed to the aqueous alteration of magnetite into ferric‐iron bearing phases. At Ernutet, Inamahari, and Urvara, the organic‐rich material is confined to the near surface only. Additionally, the absence of tectonic/volcanic features at these sites makes an endogenic origin questionable. The global rarity of detectable organics also supports this assessment. Consequently, we suggest that organics at these sites were originally delivered by low‐velocity, organic‐rich impactor(s) from the main belt and subsequently excavated, rather than originating from endogenous processes.

Design and Simulation of a Muon Detector Using Wavelength-Shifting Fiber Readouts for Border Security

Cosmic ray muon tomography is a promising method for the non-invasive inspection of shipping containers and trucks. It leverages the highly penetrating cosmic muons and their interactions with various materials to generate three-dimensional images of large and dense objects, such as inter-modal shipping containers, which are typically opaque to conventional X-ray radiography techniques. One of the key tasks of customs and border security is verifying shipping container declarations to prevent illegal trafficking, and muon tomography offers a viable solution for this purpose. Common imaging methods using muons rely on data analysis of either muon scattering or absorption–transmission. We design a compact muon tomography system with dimensions of 3 × 3 × 3 m3, consisting of 2D position-sensitive detectors. These detectors include plastic scintillators, wavelength-shifting (WLS) fibers, and SiPMs. Through light transport modeling with GEANT4, we demonstrate that the proposed detector design—featuring 1 m × 1 m scintillator plates with 2 mm2 square-shaped WLS fibers—can achieve a spatial resolution of approximately 0.7–1.0 mm. Through Monte Carlo simulations, we demonstrate that combining muon scattering and absorption data enables the rapid and accurate identification of cargo materials. In a smuggling scenario where tobacco is falsely declared as paper towel rolls, this combined analysis distinguishes the two with 3 σ confidence at a spatial resolution of 1 mm (FWHM) for the muon detector, achieving results within a scanning time of 40 s for a 20-foot shipping container.

Constructing relative spatial relationships of large area objects using time-of-flight sensing and achieving multidimensional evaluation

Constructing relative spatial relationships of large area objects using time-of-flight sensing and achieving multidimensional evaluation

“It Is All Mine, Native!”: A Sense of Place in the Context of Informal Nature Management (the Case of Karakansky Pine Wood)

The article reveals the analytical potential of the concept of a sense of place in the annex to such large natural objects as suburban forests. On the data of ethnographic case-study held in 2021-2022 in the Novosibirsk region, shown how the biophysical characteristics of the natural-anthropogenic landscape, refracting through the history of life and individual experience of natural resource management, participate in the formation of place attachment, place dependence, place identity and place satisfaction. The components of a sense of place include both rational and irrational components, voluntariness and necessity, choice and lack thereof. Personal experience of pine wood users — rural residents, summer residents and city tourists — is characterized by different degree of rootedness in the landscape, which qualitatively and quantitatively measured by the duration of permanent and seasonal residence, the presence of generational ties with the place, features of environmental management practices. The experience of place, which shapes the value of the place and through it the components of the sense of place, can be both unifying, creating unconditional value, not based on comparison (native places), and individualizing, creating comparative satisfaction and attachment to place (place resource). Strong emotional experiences of unity with nature are balanced, and often reinforced by economic difficulties, which as a result leads to perception of the place not only as the “best”, but also as “native”, “own”.

A Lightweight Small Target Detection Algorithm for UAV Platforms

The targets in the aerial view of UAVs are small, scenes are complex, and background noise is strong. Additionally, the low computational capability of UAVs is challenged when trying to meet the requirements of large neural networks. Therefore, a lightweight object detection algorithm tailored for UAV platforms, called RSG-YOLO, is proposed. The algorithm introduces an attention module constructed with receptive field attention and coordinate attention, which helps reduce background noise interference while improving long-range information dependency. It also introduces and refines a fine-grained downsampling structure to minimize the loss of target information during the downsampling process. A general efficient layer aggregation network enhances the base feature extraction module, improving gradient flow information. Additionally, a detection layer rich in small target information is added, while redundant large object detection layers are removed, achieving a lightweight design while enhancing detection accuracy. Experimental results show that, compared to the baseline algorithm, the improved algorithm increases the P, R, mAP@0.5, and mAP@0.5:0.95 by 6.9%, 7.2%, 8.4%, 5.8%, respectively, on the VisDrone 2019 dataset, and by 5.7%, 9%, 9.3%, 3.6%, respectively, on the TinyPerson dataset, while reducing the number of parameters by 23.3%. This significantly enhances the model’s detection performance and robustness, making it highly suitable for object detection tasks on low-computing-power UAV platforms.

Counterfactual rewards promote collective transport using individually controlled swarm microrobots.

Swarm robots offer fascinating opportunities to perform complex tasks beyond the capabilities of individual machines. Just as a swarm of ants collectively moves large objects, similar functions can emerge within a group of robots through individual strategies based on local sensing. However, realizing collective functions with individually controlled microrobots is particularly challenging because of their micrometer size, large number of degrees of freedom, strong thermal noise relative to the propulsion speed, and complex physical coupling between neighboring microrobots. Here, we implemented multiagent reinforcement learning (MARL) to generate a control strategy for up to 200 microrobots whose motions are individually controlled by laser spots. During the learning process, we used so-called counterfactual rewards that automatically assign credit to the individual microrobots, which allows fast and unbiased training. With the help of this efficient reward scheme, swarm microrobots learn to collectively transport a large cargo object to an arbitrary position and orientation, similar to ant swarms. We show that this flexible and versatile swarm robotic system is robust to variations in group size, the presence of malfunctioning units, and environmental noise. In addition, we let the robot swarms manipulate multiple objects simultaneously in a demonstration experiment, highlighting the benefits of distributed control and independent microrobot motion. Control strategies such as ours can potentially enable complex and automated assembly of mobile micromachines, programmable drug delivery capsules, and other advanced lab-on-a-chip applications.

Novel frequency selective B1 focusing passive lenz resonators for substantial MRI signal-to-noise ratio amplification

Objective.The need for increased sensitivity in magnetic resonance imaging (MRI) is crucial for its advancement as an imaging modality. The development of passive Lenz Resonators (LRs) for effective RF magnetic field (B1) focusing will improve MRI sensitivity via local amplification of MRI signal, thereby leading to more efficient diagnosis and patient treatment.Approach.While there are methods for amplifying the signal from specific nuclei in MRI, such as hyperpolarization, a general solution will be more advantageous and would work in combination with these preexisting methods. While the Lenz Lens proposed such a general solution based on Lenz's law and the reciprocity principle, it came at the cost of limited signal enhancement. In this work, the first-in-kind prototype LR was conceived and examined as a general frequency-selective passive flux-focusing element for significant MRI signal enhancement. A 3.0 T Philips Achieva MRI was used to compare the signal from a phantom in the presence of Lenz Lenses, LRs, and control trials with neither component.Main results.An MRI investigation demonstrated an experimental amplification of the signal-to-noise ratio up to 80% using an MRI insert of two coaxial LRs due to superior B1focusing. The resonators displayed consistent amplification, nearly independent of theirx-position within the MRI bore.Significance.This behavior demonstrates the feasibility of imaging large objects of varying shapes without penalties for signal amplification using LRs. The LRs versatility in geometrical design and consistent signal amplifying abilities between pulse sequences should allow for the development of LRs suitable for most commonly used MRI setups.

Analysis of Natural and Technological Losses of Products in the Collection, Preliminary Preparation and Storage Systems at the Kokdumalok Umid and South Kemachi Mines

This article examines and analyzes the collection and preparation systems in oil fields. It explores the conditions under which various technological schemes are applied during the process. The establishment of large centralized collection points greatly simplifies the collection schemes of oil and liquid hydrocarbons from separate deposits and creates favorable conditions for their integration into large objects. It is more profitable to separate oil and gas in large centralized facilities than in small facilities. Such centralization allows to reduce the loss of light oil fractions, to improve oil production, to provide deeper processing of gas, and to ensure maximum extraction of raw materials for the chemical industry. Today, in the oil and gas regions of many countries, great work is being done on the development of oil and gas gathering systems, and the technical and economic basis of oil and gas gathering and preparation and shipping is being analyzed. In the future, the development of optimized oil collection and preparation systems is planned, with the goal of integrating these systems into national budgets. This article was prepared on the example of a mine as part of the research and includes the application of the results in these mines. The technological losses in the fields during the initial preparation of the extracted oil in the fields were analyzed. In this process, losses of oil with associated gas, together with separated water and by evaporation in the reservoir are considered. The article also calculates the amount of oil that could be recovered by mixing with satellite gas, specifically for the Kokdumaloq and Southern Kemachi fields, and provides recommendations for reducing these losses.

BRA-YOLOv7: improvements on large leaf disease object detection using FasterNet and dual-level routing attention in YOLOv7

Tea leaf diseases are significant causes of reduced quality and yield in tea production. In the Yunnan region, where the climate is suitable for tea cultivation, tea leaf diseases are small, scattered, and vary in scale, making their detection challenging due to complex backgrounds and issues such as occlusion, overlap, and lighting variations. Existing object detection models often struggle to achieve high accuracy in detecting tea leaf diseases. To address these challenges, this paper proposes a tea leaf disease detection model, BRA-YOLOv7, which combines a dual-level routing dynamic sparse attention mechanism for fast identification of tea leaf diseases in complex scenarios. BRA-YOLOv7 incorporates PConv and FasterNet as replacements for the original network structure of YOLOv7, reducing the number of floating-point operations and improving efficiency. In the Neck layer, a dual-level routing dynamic sparse attention mechanism is introduced to enable flexible computation allocation and content awareness, enhancing the model’s ability to capture global information about tea leaf diseases. Finally, the loss function is replaced with MPDIoU to enhance target localization accuracy and reduce false detection cases. Experiments and analysis were conducted on a collected dataset using the Faster R-CNN, YOLOv6, and YOLOv7 models, with Mean Average Precision (mAP), Floating-point Operations (FLOPs), and Frames Per Second (FPS) as evaluation metrics for accuracy and efficiency. The experimental results show that the improved algorithm achieved a 4.8% improvement in recognition accuracy, a 5.3% improvement in recall rate, a 5% improvement in balance score, and a 2.6% improvement in mAP compared to the traditional YOLOv7 algorithm. Furthermore, in external validation, the floating-point operation count decreased by 1.4G, FPS improved by 5.52%, and mAP increased by 2.4%. In conclusion, the improved YOLOv7 model demonstrates remarkable results in terms of parameter quantity, floating-point operation count, model size, and convergence time. It provides efficient lossless identification while balancing recognition accuracy, real-time performance, and model robustness. This has significant implications for adopting targeted preventive measures against tea leaf diseases in the future.

Construction of tennis pose estimation and action recognition model based on improved ST-GCN

With the rapid growth of computer vision and deep learning technologies, the application of pose estimation and action recognition in sports training has become increasingly widespread. Due to factors such as complex movements, fast speed, and limb occlusion, pose estimation and action recognition in tennis face significant challenges. Therefore, this study first introduces selective dropout and pyramid region of interest pooling layer strategies in fast region convolutional neural networks. Secondly, a pose estimation algorithm based on multi-scale fusion pose residual network 50 is designed, and finally a spatiotemporal graph convolutional network model is constructed by fusing channel attention module and multi-scale dilated convolution module. The data showed that the average detection accuracy of the improved attitude residual network 50 was 70.4%, and the accuracy of object detection for small, medium, and large objects was 57.4%, 69.3%, and 79.2%, respectively. The continuous action recognition accuracy and inter action fluency detection time of the improved spatiotemporal graph convolutional network were 93.8% and 19.2 ms, respectively. When the sample size was 1000, its memory usage was 1378 MB and the running time was 32.7 ms. Experiments have shown that the improved model achieves high accuracy and robustness in tennis action recognition tasks, especially in complex scenes and limb occlusion conditions, where the model shows significant advantages. This study aims to provide an efficient and accurate motion recognition technology for tennis posture analysis and intelligent training.

Cooperative dual-actor proximal policy optimization algorithm for multi-robot complex control task

This paper introduces a novel multi-agent Deep Reinforcement Learning (DRL) framework named the Cooperative Dual-Actor Proximal Policy Optimization (CDA-PPO) algorithm, designed to address complex humanoid robot cooperative learning control tasks. Effective cooperation among multiple humanoid robots, particularly in scenarios involving complex walking gait control and external disturbances in dynamic environments, is a critical challenge. This is especially pertinent for tasks requiring precise coordination and control, such as joint object transportation. In various real-life scenarios, humanoid robots might need to cooperate to carry large objects in many scenarios. This capability is crucial for logistics, manufacturing, intelligent transportation, and search-and-rescue missions applications. Humanoid robots have gained significant popularity, and their use in these cooperative tasks is becoming more common. To address this challenge, we propose CDA-PPO, which introduces a learning-based communication platform between agents and employs two distinct policy networks for each agent. This dual-policy approach enhances the robots’ ability to adapt to complex interactions and maintain stability while performing intricate tasks. We demonstrate the efficacy of CDA-PPO in a cooperative object-transportation scenario, where two humanoid robots collaborate to carry a table. The experimental results show that CDA-PPO significantly outperforms traditional methods, such as Independent PPO (IPPO), Multi-Agent PPO (MAPPO), and Multi-Agent Twin Delayed Deep Deterministic Policy Gradient (MATD3), in terms of training efficiency, stability, reward acquisition, and humanoid robot cooperative balance control with effective coordination between robots. The findings underscore the potential of CDA-PPO to advance the field of cooperative multi-agent control problems, proposing the way for future research in complex robotics applications.

Integrating collaborative robots in manufacturing, logistics, and agriculture: Expert perspectives on technical, safety, and human factors.

This study investigates the implementation of collaborative robots across three distinct industrial sectors: vehicle assembly, warehouse logistics, and agricultural operations. Through the SESTOSENSO project, an EU-funded initiative, we examined expert perspectives on human-robot collaboration using a mixed-methods approach. Data were collected from 31 technical experts across nine European countries through an online questionnaire combining qualitative assessments of specific use cases and quantitative measures of attitudes, trust, and safety perceptions. Expert opinions across the use cases emphasized three primary concerns: technical impacts of cobot adoption, social and ethical considerations, and safety issues in design and deployment. In vehicle assembly, experts stressed the importance of effective collaboration between cobots and exoskeletons to predict and prevent collisions. For logistics, they highlighted the need for adaptable systems capable of handling various object sizes while maintaining worker safety. In agricultural settings, experts emphasized the importance of developing inherently safe applications that can operate effectively on uneven terrain while reducing workers' physical strain. Results reveal sector-specific challenges and opportunities: vehicle assembly operations require sophisticated sensor systems for cobot-exoskeleton integration; warehouse logistics demand advanced control systems for large object handling; and agricultural applications need robust navigation systems for uneven terrain. Quantitative findings indicate generally positive attitudes toward cobots, particularly regarding societal benefits, moderate to high levels of trust in cobot capabilities and favorable safety perceptions. The study highlights three key implications: (1) the need for comprehensive safety protocols tailored to each sector's unique requirements, (2) the importance of user-friendly interfaces and intuitive programming methods for successful cobot integration, and (3) the necessity of addressing workforce transition and skill development concerns. These findings contribute to our understanding of human-robot collaboration in industrial settings and provide practical guidance for organizations implementing collaborative robotics while considering both technological advancement and human-centered design principles.

Collaborative Bimanual Manipulation Using Optimal Motion Adaptation and Interaction Control: Retargeting Human Commands to Feasible Robot Control References

This article presents a robust and reliable human–robot collaboration (HRC) framework for bimanual manipulation. We propose an optimal motion adaptation method to retarget arbitrary human commands to feasible robot pose references while maintaining payload stability. The framework comprises three modules: 1) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a task-space sequential equilibrium and inverse kinematics optimization</i> ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">task-space SEIKO</i> ) for retargeting human commands and enforcing feasibility constraints, 2) an admittance controller to facilitate compliant human–robot physical interactions, and 3) a low-level controller improving stability during physical interactions. Experimental results show that the proposed framework successfully adapted infeasible and dangerous human commands into continuous motions within safe boundaries and achieved stable grasping and maneuvering of large and heavy objects on a real dual-arm robot via teleoperation and physical interaction. Furthermore, the framework demonstrated the capability in the assembly task of building blocks and the insertion task of industrial power connectors.

Spatial-Temporal Variations and Severity of the 2020 Catastrophic Floods in the Yangtze River Basin from Sentinel-1 SAR Data

Flood is one of the most frequent natural disasters in the Yangtze River Basin. Flood risk evaluation is of great social significance, especially for large hydrological systems. Rainfall is both temporal and spatial, influencing surface hydrological activities. The water body range is the final outcome of a flood and can be observed from synthetic aperture radar (SAR) images under any weather condition. A flood severity evaluation model is proposed to quantitatively evaluate the flood based on water body range from area disparity and flood duration. Large hydrological objects usually span a wide range and have significant differences. This results in different initial water areas in each region. This approach addresses the issue through normalization processing. In this paper, Sentinel-1 data are used to extract the temporal water body using the adaptive bimodal method, and the water level data were also incorporated to improve the observation frequency for water area. The flood severity evaluation approach can be used to assess flood risk between any region of large hydrological systems or any flood event, regardless of their regional spatial differences and rainfall duration differences. The results show that: (1) In general, the average water body area in 2020 was 20.40% larger than it was in 2019, and the daily water body areas in 2020 were all greater than the average of 2019 with 71.36% of the days in 2020 having an area greater than the maximum in 2019. The flood severity in 2020 was 1.75 times as much as that of 2019; (2) Reach performance indexes in 2020 were in order of Yueyang (2.21) &gt; Jiujiang (2.04) &gt; Hankou (1.44) &gt; Chizhou (1.32), which were inconsistent with the spatial site; (3) Flood event impact indexes in 2020 were in order as No.2 (1.64) &gt; No.3 (1.61) &gt; No.1 (1.44) &gt; No.4 (1.17) &gt; No.5 (1.15); (4) The flood was more likely the result of cumulative rainfall for 30 days.

HawkEye Conv-Driven YOLOv10 with Advanced Feature Pyramid Networks for Small Object Detection in UAV Imagery

Current mainstream computer vision algorithms focus on designing suitable network architectures and loss functions to fit training data. However, the accuracy of small object detection remains lower than for other scales, and the design of convolution operators limits the model’s performance. For UAV small object detection, standard convolutions, due to their fixed kernel size, cannot adaptively capture small object spatial information. Many convolutional variants have scattered sampling points, leading to blurred boundaries and reduced accuracy. In response, we propose HawkEye Conv (HEConv), which utilizes stable sampling and dynamic offsets with random selection. By varying the convolution kernel design, HEConv reduces the accuracy gap between small and larger objects while offering multiple versions and plug-and-play capabilities. We also develop HawkEye Spatial Pyramid Pooling and Gradual Dynamic Feature Pyramid Network modules to validate HEConv. Experiments on the RFRB agricultural and VisDrone2019 urban datasets demonstrate that, compared to YOLOv10, our model improves AP50 by 11.9% and 6.2%, APS by 11.5% and 5%, and F1-score by 5% and 7%. Importantly, it enhances small object detection without sacrificing large object accuracy, thereby reducing the multi-scale performance gap.

Real scale experiments on the wave-induced burial and mobilization of Unexploded Ordnance on the seafloor

As a result of armed conflicts, huge amounts of Unexploded Ordnance devices (UXO) and Discarded Munition Material (DMM) are expected to be located on the seafloor, especially in coastal regions. During Offshore Construction, strategical site monitoring and systematic remediation activities, the behavior of such objects in waves and currents is of huge interest as potential mobilization of objects after a survey or clearance activity could change the situational picture again. From findings and reports it often is assumed that UXO and DMM tend to migrate over the seafloor for long distances. Here, anthropogenic effects like fishing or dredging activities are underestimated. More scientific approaches clearly show that mobilization and migration of UXO over long distances does not occur. However, theoretical analysis remain theories until they are proven by experiments. For this reason, three large representative objects were investigated under real scale wave conditions in the Delta Flume of Deltares. The objects represent real scale models of UXO found in the North Sea as well as academically shaped objects. All objects as well as the Flume Tank were intensively instrumented to measure the full environmental conditions as well as the behavior of the objects. The sediment represents typical sand as found in the North Sea and the seabed morphology and soil conditions were closely monitored during the experiments. The experiments support the theoretical models that predict burial but no mobilization also under extreme wave conditions.

Radiographic Inspection of Carbon Fiber-Reinforced Polymer Composites (Laminates) with Epoxy and PEEK Binders After Impact and Subsequent Compression Loading.

An approach to detecting discontinuities in carbon fiber-reinforced polymers, caused by impact loading followed by compression testing, was developed. An X-ray sensor-based installation was used, while some algorithms were developed to improve the quality of the obtained low-contrast radiographic images with negligible signal-to-noise ratios. For epoxy/AF (#1) composite subjected to a "high-velocity" steel-ball impact with subsequent compression loading, it was not possible to detect discontinuities since the orientation of the extended zone of interlayer delamination was perpendicular to the irradiation axis. After drop-weight impacts with subsequent compression loading of epoxy/CF (#2) and PEEK/CF (#3) composites, the main cracks were formed in their central parts. This area was reliably detected through the improved radiographic images being more contrasted compared to that for composite #3, for which the damaged area was similar in shape but smaller. The phase variation and congruency methods were employed to highlight low-contrast objects in the radiographic images. The phase variation procedure showed higher efficiency in detecting small objects, while phase congruency is preferable for highlighting large objects. To assess the degree of image improvement, several metrics were implemented. In the analysis of the model images, the most indicative was the PSNR parameter (with a S-N ratio greater than the unit), confirming an increase in image contrast and a decrease in noise level. The NIQE and PIQE parameters enabled the correct assessment of image quality even with the S-N ratio being less than a unit.

The Lightweight Fracture Segmentation Algorithm for Logging Images Based on Fully 3D Attention Mechanism and Deformable Convolution

The challenge of fracture segmentation remains a significant obstacle in imaging logging interpretation within the current oil and gas exploration and development field. However, existing image segmentation algorithms still encounter issues related to accuracy, speed, and robustness, as well as a tendency to misdetect or overlook small fractures when applied to logging image fracture segmentation tasks. To address these challenges comprehensively, this paper proposes an end-to-end fracture segmentation algorithm named SWSDS-Net. This algorithm is built upon the UNet architecture and incorporates the SimAM with slicing (SWS) attention mechanism along with the deformable strip convolution (DSCN) module. The SWS introduces a fully 3D attention mechanism that effectively learns the weights of each neuron in the feature map, enabling better capture of fracture features while ensuring fair attention and enhancement for both large and small objects. Additionally, the deformable properties of DSCN allow for adaptive sampling based on fracture shapes, effectively tackling challenges posed by varying fracture shapes and enhancing segmentation robustness. Experimental results demonstrate that SWSDS-Net achieves optimal performance across all evaluation metrics in this task, delivering superior visual results in fracture segmentation while successfully overcoming limitations present in existing algorithms such as complex shapes, noise interference, and low-quality images. Moreover, serving as a lightweight network solution enables SWSDS-Net’s deployment on mobile devices at remote sites—an advancement that lays a solid foundation for interpreting logging data and promotes deep learning technology application within traditional industrial scenarios.

Enhancing video frame interpolation with region of motion loss and self-attention mechanisms: A dual approach to address large, nonlinear motions

Video frame interpolation is particularly challenging when dealing with large and non-linear object motions, often resulting in poor frame quality and motion artifacts. In this study, we introduce a novel dual-approach methodology for video frame interpolation that effectively addresses these complexities. Our method consists of two key components: a Region of Motion (RoM) loss and self-attention mechanisms. The RoM loss is designed to spotlight significant movements within frames. This is achieved by employing feature-matching techniques that assign tailored weights during the training process, ensuring that areas of intense motion are given priority. This is facilitated by the computation of optical flow, which identifies crucial feature points and highlights regions of significant motion for targeted enhancement. Our method incorporates self-attention mechanisms to maintain inter-frame continuity while emphasizing the unique attributes of individual frames. The self-attention scores reduce motion discrepancies and enhance the distinctiveness and texture quality of each frame. We validate the efficacy of our approach through extensive evaluations on benchmark datasets, including Vimeo-90K, Middlebury, UCF101, and SNU-Film.