Diverse Shapes Research Articles

Res2U++: Deep learning model for segmentation of ischemic stroke lesions

Stroke is a critical global health issue characterized by the interruption of blood flow to the brain, resulting in a depletion of oxygen and nutrients and causing severe damage or death to brain cells. Early detection and intervention are pivotal in preventing adverse outcomes. Diagnostic imaging modalities like CT and MRI are essential for accurate diagnosis and treatment planning. However, precise segmentation of stroke lesions remains challenging due to variations in appearance, size, location, and intensity within the brain, exacerbated by noise and variability in manual segmentation. To address these challenges, we propose a novel deep neural network model for automatic stroke segmentation. Our model features an encoder–decoder architecture, combining the robust deep and multi-scale feature extraction capabilities of the ResUnet++ encoder with a modified UNet decoder that enhances segmentation accuracy and preserves spatial information effectively. Incorporating the Atrous Spatial Pyramidal Pooling (ASPP) block in the bottleneck strengthens the model, enabling efficient capture of multi-scale contextual information and improving its capability to handle diverse object sizes and shapes in semantic segmentation tasks. We evaluate the model on the Ischemic Stroke Lesion Segmentation Challenge 2015 (SISS 2015) and ISLES 2022 databases. During training, our model achieves a Dice coefficient of 0.85 ± 0.18 on SISS 2015 and 0.85 ± 0.11 on ISLES 2022. For testing on SISS 2015, it achieves 0.80 ± 0.33, surpassing current state-of-the-art results for these datasets.

Factors Diversifying the Characteristics of Fluvial Sediments Accumulated in Mountain Stream Channels—A Case Study from the Polish Carpathians

This paper presents the diversification of fluvial sediments caused by the occurrence of coarse woody debris (CWD), boulder steps (BSs), and mixed structures (MSs), understood as a combination of CWD and BSs in a stream channel in a small forested catchment in the Polish Carpathians. This research is crucial for understanding the role of this kind of threshold present in a stream channel in shaping fluvial sediment characteristics in small forested mountain catchments. Our hypothesis is that the threshold type in a stream channel determines fluvial sediment diversification. This was verified in field research, including identification of the channel’s morphodynamic structure and the morphometric characteristics of CWD, BSs, and MSs as well as the collection of fluvial sediments upstream and downstream of them. In order to preserve research objectivity, tests were performed during comparable flow conditions in the summer (EX1) and autumn (EX2) periods. The statistical analysis showed that the type of threshold significantly affects the processing, size, and shape diversification of mineral material. This diversity is particularly noticeable in fluvial sediments within CWD and MSs, which retain material of more diverse sizes and shapes.

Primera aproximación a los artefactos de molienda del sudeste del lago del Colhué Huapi (Chubut, Argentina)

The Sarmiento Basin (Chubut, Argentina) has offered a privileged environment for the development of human life over the Holocene. For decades it was considered that the groups that inhabited this area based their economy on the hunting of terrestrial mammals, mainly Lama guanicoe. However, the new evidence suggests a diversification in the economy at particular times, adding new resources, technologies and/or ways of processing them. Within this framework, the goal of this work was to carry out a first characterization of grinding artifacts found at the southeast of the Colhué Huapi lake. For this purpose, distributional and spatial arrangement were analyzed, as well as dimensional, state of conservation and general techno-morphological variables. The results warn of a high number of active and passive artifacts made with local raw materials, whose diversity of shapes, weights, and states of conservation suggest a heavy equipment of space. Added to other technologies —such as ceramics and specific artifacts for fishing—, they would indicate an intensification in the use of resources towards the end of the late Holocene, suggesting as well an important place for grinding practices in the life of the hunter-gatherers who inhabited this area.

Extreme wrinkling of the nuclear lamina is a morphological marker of cancer

Nuclear atypia is a hallmark of cancer. A recent model posits that excess surface area, visible as folds/wrinkles in the lamina of a rounded nucleus, allows the nucleus to take on diverse shapes with little mechanical resistance. Whether this model is applicable to normal and cancer nuclei in human tissues is unclear. We image nuclear lamins in patient tissues and find: (a) nuclear laminar wrinkles are present in control and cancer tissue but are obscured in hematoxylin and eosin (H&E) images, (b) nuclei rarely have a smooth lamina, and (c) wrinkled nuclei assume diverse shapes. Deep learning reveals the presence of extreme nuclear laminar wrinkling in cancer tissues, which is confirmed by Fourier analysis. These data support a model in which excess surface area in the nuclear lamina enables nuclear shape diversity in vivo. Extreme laminar wrinkling is a marker of cancer, and imaging the lamina may benefit cancer diagnosis.

CylinderTag: An Accurate and Flexible Marker for Cylinder-Shape Objects Pose Estimation Based on Projective Invariants.

High-precision pose estimation based on visual markers has been a thriving research topic in the field of computer vision. However, the suitability of traditional flat markers on curved objects is limited due to the diverse shapes of curved surfaces, which hinders the development of high-precision pose estimation for curved objects. Therefore, this paper proposes a novel visual marker called CylinderTag, which is designed for developable curved surfaces such as cylindrical surfaces. CylinderTag is a cyclic marker that can be firmly attached to objects with a cylindrical shape. Leveraging the manifold assumption, the cross-ratio in projective invariance is utilized for encoding in the direction of zero curvature on the surface. Additionally, to facilitate the usage of CylinderTag, we propose a heuristic search-based marker generator and a high-performance recognizer as well. Moreover, an all-encompassing evaluation of CylinderTag properties is conducted by means of extensive experimentation, covering detection rate, detection speed, dictionary size, localization jitter, and pose estimation accuracy. CylinderTag showcases superior detection performance from varying view angles in comparison to traditional visual markers, accompanied by higher localization accuracy. Furthermore, CylinderTag boasts real-time detection capability and an extensive marker dictionary, offering enhanced versatility and practicality in a wide range of applications. Experimental results demonstrate that the CylinderTag is a highly promising visual marker for use on cylindrical-like surfaces, thus offering important guidance for future research on high-precision visual localization of cylinder-shaped objects.

Priority directions of onion breeding (Allium cepa L.)

Relevance. Onion (Allium cepa L.) is a valuable and demanded vegetable crop, which occupies one of the leading places in terms of sown areas and gross yields among other vegetable crops of the genus Allium L. Onion is popular in folk and evidence-based medicine, widely used for the prevention and treatment of many diseases due to its rich biochemical composition and useful properties. As part of import substitution and achieving food security of the country, scientists of the Federal State Budgetary Scientific Institution «Federal Scientific Vegetable Center» (FSBSI FSVC) are working intensively to create new generation varieties and hybrids.Methodology. Various resources on the topic were selected, publications in scientific and industry journals and databases were studied, a review of works on onion selection (yield, species diversity in shape and color, maturity, shelf life, biochemical parameters, commercial and seed productivity, resistance to biotic and abiotic stresses), as well as useful properties for humans and use in medicine was made.Results. The analysis of works shows that the wide distribution in the Russian Federation have such varieties of selection of FSBSI FSVC, such as Myachkovskij 300, Cherni prince, Al'ba, Atas, Ampeks and others. In recent years, created a medium-maturing hybrid onion onion F1 Drakon with brown dry scales and a maximum yield of 391.0 c/ha, as well as transferred to the State variety trial winter variety Novator with brown dry scales. Propagation of demanded varieties and hybrids for both production cultivation and for use in household farms is underway.Conclusion. In order to conveyor onion production, it is necessary to create varieties and hybrids of different maturity groups, as well as to work on improving marketability, seed productivity and resistance to various stresses.

Biofouling changes the settling dynamics of macroplastic plates

Plastic pollution transported in rivers remains poorly understood due to the diversity of shapes, sizes, and densities of plastics, as well as their complex interactions with biofilms. While previous studies have explored the settling velocities of plastics and their interactions with biofilms, they often overlook how biofouling alters plastic dynamics and settling behaviour. To address this, over 800 settling experiments were conducted to demonstrate that the dynamics and falling velocities of isotropic (spheres) and anisotropic (square and rectangle plates) macroplastics of different densities (1050 to 2200 kg/m3) are significantly impacted by biofouling. Three-dimensional tracking of plastic trajectories revealed that biofilm colonisation on the surface of anisotropic plastic plates triggered them to exhibit more chaotic trajectories, larger horizontal dispersion and higher oscillatory frequencies. These dynamics reduced the average vertical settling velocity of anisotropic biofouled plates by up to 12%—despite greater plastic densities and considering the multimodal distribution of a plate’s fall velocity—compared to their pristine counterparts. Results highlight the necessity of accounting for the intricate multimodal settling dynamics of plastics, including their interactions with biofilms, to provide more reliable predictions of plastic transport and fate in aquatic environments.

Multi-objective optimization of horizontal louver systems with flat, single-curvature, and double-curvature profiles to enhance daylighting, glare control, and energy consumption in office buildings

Identifying the most optimal slat shape significantly influences the performance of louver systems in terms of daylighting, glare control, and energy consumption. This is particularly crucial in climates with high levels of solar irradiance, where thermal gain and daylight illumination highly affect buildings and occupants. This study aims to identify the optimal slat shapes for various profile types—flat, single-curvature, and double-curvature—that simultaneously reduce annual energy consumption and enhance annual daylighting and visual comfort performance in fully glazed office buildings. This goal is achieved by employing a multi-objective optimization (MOO) algorithm, NSGA-II, applied to a control-point-based algorithm designed to generate diverse slat shapes for each profile type. The main objective functions include the maximization of spatial useful daylight illuminance (sUDI) and spatial glare autonomy (sGA), as well as the minimization of energy use intensity (EUI). The MOO process results in a diverse set of Pareto optimal slat shapes for each profile type, which are subsequently ranked by a fitness function. Findings suggest that the Pareto optimal solutions within each type significantly improve the overall performance of the space compared to the base case. Specifically, among these solutions, flat profiles with the highest fitness scores enhance daylighting levels of the space to a greater extent (9.028% to 14.583%) compared to single (−2.778% to 12.5%) and double-curvature profiles (−5.556% to 9.722%) with the highest scores. Regarding glare, double-curvature profiles with the highest fitness scores provide a more visually comfortable environment for users by improving the sGA value by 19.879% to 33.247% compared to the base case. However, those with a concave-convex shape produce excessive illumination in the perimeter zone, whereas those with a convex-concave shape present challenges in providing sufficient daylight in the rear zone of the space. Additionally, applying any of the Pareto optimal solutions to the louver system reduces the annual energy consumption of the office space compared to the base model. While the improvements are almost comparable, flat profiles with the highest fitness scores marginally reduces energy consumption to a greater extent (62.906% to 68.161%).

Hybrid dynamically cross-linked polyurethanes with room temperature self-healing ability for light-responsive actuator

Light-responsive actuators (LRAs) with cross-linked networks that demonstrate fast response and super mechanical strength are essential for practical applications. However, the permanent cross-linked structure typically stabilizes polymer networks and makes them difficult to reprogram or self-heal. Herein, a novel crosslinked light-responsive polyurethane actuator (CPU-DA1SS2) with room temperature self-healing properties was prepared by constructing a hybrid physical and chemical cross-linked polymer network via dynamic DA bonds, SS bonds, and hydrogen bonds. The CPU-DA1SS2 light-responsive actuator exhibits high tensile strength (25.52 MPa) and excellent self-healing efficiency (90.16 %) at room temperature. Moreover, the hybrid dynamic cross-linked network endows the LRAs with liquid-state remold ability and solid-state plasticity, which further contributes to diverse 3D geometric shape remolding. Such a hybrid, dynamic, cross-linked strategy could be utilized to design smart materials that are self-healing and recyclable without sacrificing their mechanical properties, helping to broaden the application areas of LRA materials.

Versatile unsupervised design of antennas using flexible parameterization and computational intelligence methods

Developing contemporary antennas is a challenging endeavor that requires considerable engineering insight. The most laborious stage is to devise an antenna architecture that delivers the required functionalities, e.g., multiband operation. Iterative by nature (hands-on topology modifications, parametric studies, trial-and-error geometry selection), it typically takes many weeks and requires considerable engagement from a human expert. Consequently, only a few possible design options concerning the fundamental antenna geometry may be considered. Automated topology rendition and geometry parameter optimization are highly relevant, especially from the industrial perspective. Therein, reducing time-to-market and limiting the involvement of trained experts is critical. This research proposes an innovative procedure for unsupervised development of planar antennas. Our method leverages flexible antenna parameterization based on re-sizable elliptical patches. It permits the realization of a massive number of geometries of diverse shapes and complexities using a small number of decision variables. Computational intelligence methods are employed to conduct antenna evolution exclusively based on specifications and possible constraints (e.g., maximum size). Fine-tuning of the structure geometry is achieved through low-cost local search routines. Our methodology is demonstrated by designing several antennas featuring distinct characteristics (broadband, single-, dual- and triple-band). The obtained results, supported by experimental data, underscore the presented approach’s versatility and capability to render unconventional topologies at reasonably low computational expenses. As mentioned earlier, the design process is fully automated without human expert involvement.

Photorealistic attention style transfer network for architectural photography photos

Architectural photography style transfer, a task in computer vision, employs deep learning algorithms to transform the style of architectural photograph while preserving key structure and content. Existing algorithms face challenges due to the intricate details of buildings, including diverse shapes, lines, and textures. Moreover, considerations for artistic effects in architectural photography style transfer, such as lighting, shadows, and atmosphere, require high-quality image generation algorithms. However, current algorithms often struggle to address these complexities, leading to loss or blurring of details and less realistic images. To overcome these challenges, this paper proposes a Photorealistic Attention Style Transfer Network. The proposed approach utilizes a semantic segmentation model to accurately segment the input image into foreground and background components for independent style transfer. Subsequently, the transferred images are refined by focusing on intricate building parts using the coordinate attention mechanism. Additionally, the network incorporates loss functions to capture light, shadow, and colors in stylish images, ensuring realism while maintaining aesthetic appeal. Through comparative experiments, the proposed network shows better performance in terms of image fidelity and overall aesthetics, and the SSIM and PSNR indices are also better than the current mainstream methods.

The miR319-based repression of SlTCP2/LANCEOLATE activity is required for regulating tomato fruit shape.

Fruit morphogenesis is determined by the coordination of cell division and expansion, which are fundamental processes required for the development of all plant organs. Here, we show that the regulation of TEOSINTE BRANCHED1/CYCLOIDEA/PCF (TCP) LANCEOLATE (TCP2/LA) by miR319 is crucial for tomato fruit morphology. The loss of miR319 regulation in the semi-dominant La mutant led to a premature SlTCP2/LA expression during gynoecium patterning, which results in modified cell division during carpel development. As a consequence, La mutants exhibited elongated ovary and fruit shape, and a reduced number of ovules and seeds. Elongated fruit shape in La may be partially due to the SlTCP2/LA-mediated repression of OVATE activity in young floral buds. Further analysis showed that the de-repression of SlTCP2/LA decreases auxin responses in young floral buds by directly repressing SlYUCCA4 expression, but SlTCP2/LA also acts in parallel with ENTIRE (E) to orchestrate fruit morphology and seed production. Our study defines a novel miRNA-based molecular link between the domestication-associated OVATE gene and auxin responses. Given the striking variation in fruit morphology among members of the Solanaceae family, fine-tuning regulation of gene expression by miRNA coupled with modulation of auxin dynamics may be a common driver in the evolution of fruit shape diversity.

Complex landscape, instream and anthropogenic environmental features drive genetic and morphological structuring amongst brown trout (Salmo trutta) populations in a dendritic river system

Abstract Untangling the patterns and proximate drivers of intraspecific genetic and phenotypic structuring informs our understanding of the evolutionary processes shaping diversity. This study investigated morphological and genetic structuring of brown trout (Salmo trutta L.) populations across varying spatial scales in a single, complex, dendritic river catchment and examined the potential natural and anthropogenic environmental features driving this structuring. Morphometric and hierarchical genetic structuring analyses of fish from 22 sampling sites in the River Foyle catchment, Ireland (~ 4500km2) identified 19 morphologically distinct groups and 15 genetically distinct populations, separated by river distances ranging from 0.4 km to 188 km. Isolation by Distance was the main factor shaping both genetic and morphological divergence, indicating that strong philopatry is one of the major drivers of the observed population structuring in this system. However, both natural and anthropogenic environmental variables also explained pairwise genetic and morphological differences between sampling sites. Thus, the pairwise differences in the area of woodland in the upstream catchment, water phosphorus concentration, biological oxygen demand, catchment slope, urban area in upstream catchment, altitude, site specific percentage of canopy cover and dissolved oxygen concentration in the river channel were correlated with genetic divergence. The pairwise differences in the concentration of suspended solids, the extent of bankside overhang, the composition of bedrock, boulder and cobble substrates, watercourse width, catchment slope and site altitude were correlated with between-site morphological differences. We hypothesise that local differential selection pressures comprising both natural environmental variation and variation resulting from anthropogenic effects, in combination with strong philopatry and random genetic processes drive the clearly defined genetic and phenotypic patterns described here.

A multimodal framework for pepper diseases and pests detection

Pepper diseases and pests typically exhibit small target proportions, diverse shapes and sizes, complex imaging backgrounds, and similarities with the background. Existing detection methods perform poorly in identifying targets of different sizes and shapes within the same scene, and they lack adequate noise suppression capabilities. To address the practical needs of detecting pepper diseases and pests in complex scenarios, we have constructed the first multimodal pepper diseases and pests object detection dataset (PDD). This dataset includes a wide variety of diseases and pests images, along with detailed natural language descriptions of their attributes. Locating the described targets in complex scenes with similar disease symptoms and leaf occlusion presents a significant challenge. To tackle this issue, we propose the PepperNet model for object detection in pepper diseases and pests images using natural language descriptions. This model decomposes complex multimodal features of language and images into explicit attribute features and employs fine-grained multimodal attribute contrast learning strategies. This approach effectively distinguishes subtle local differences between similar objects, achieving fine-grained mapping from language to vision in complex scenarios. Our detection results show a mAP@0.5 of 91.93% and a detection speed of 121.8 frames per second. Visualizations indicate that the model maintains high robustness under varying noise levels and occlusion conditions, demonstrating superior performance and stability across diverse complex scenarios.

Regulatory mechanism and applicability of pulsed gas on drying uniformity of non-spherical wet grains in a fluidized-bed

Pulsed gas can regulate the uniform fluidization of particles, thereby enhancing the product quality of wet grains after drying. However, due to the diversity of grain shapes, the regulatory mechanism and applicability of pulsed gas on the drying uniformity of grains remain unclear. This study used an improved CFD-DEM drying model for wet grains to investigate how particle shape and pulsed frequency affect flow, heat transfer, and mass transfer. The results showed that increasing pulsed frequency can transition the particle mixing mechanism from convective to diffusive, and reduce the particle residence time in high-temperature regions. The formation frequency of horizontal channel-like bubbles influences the mesoscale structure evolution and interphase heat transfer uniformity. However, higher pulsed frequency decreases contact time between particles and bubbles, requiring a balance between mixing intensity and bubble contact. Finally, the intercriteria correlation method quantified the applicability of pulsed frequency for drying particles with different shapes.

Cellular mechanism of polarized auxin transport on fruit shape determination revealed by time-lapse live imaging.

Polarized auxin transport regulates fruit shape determination by promoting anisotropic cell growth. Angiosperms produce organs with distinct shape resultant from adaptive evolution. Understanding the cellular basis underlying the development of plant organ has been a central topic in plant biology as it is key to unlock the mechanisms leading to the diversification of plants. Variations in the location of synthesis, polarized auxin transport (PAT) have been proposed to account for the development of diverse organ shapes, but the exact cellular mechanism has yet to be elucidated. The Capsella rubella develops a perfect heart-shaped fruit from an ovate shape gynoecium that is tightly linked to the localized auxin synthesis in the valve tips and provides a unique opportunity to address this question. In this study, we studied auxin movement in the fruits and the cellular effect of N-1-Naphthylphthalamic Acid (NPA) on the fruitshape determination by constructing the pCrPIN3:PIN3:GFP reporter and live-imaging. We found PAT in the valve epidermis is in congruent with fruit shape development and NPA treatment disrupts the heat-shaped fruit development mainly by repressing cell anisotropic growth with minor effect on division. As the Capsella fruit is unusually big in size, we also included a detailed step-by-step protocol on how to conduct live-imaging experiment. We further test the utility of this protocol by conducting a live-imaging analysis of the gynophore in Arachis hypogaea. Collectively, the results of this study elucidated the mechanism on how auxin signal was translated into instructions guiding cell growth during organ shape determination. In addition, the description of the detailed live-imaging protocol will encourage further studies of the cellular mechanisms underlying shape diversification in angiosperms.

Linguistic diversity shapes flexible speech perception in school age children

Every day, listeners encounter a wide range of acoustic signals. Successfully solving this variability problem allows them to interpret these signals accurately. While this mechanism tends to be less effortful for adults, children need to learn stable categories in the face of such variability. It is unknown to what extent general maturation or diversity of the input plays a role in shaping different speech categorization profiles that children can employ. Here, we tested school-aged children’s speech categorization with a continuous speech categorization task called the Visual Analogue Scaling (VAS) task. We measured the linguistic diversity in each child’s social environment through a social network analysis. We found that increased linguistic diversity led to more flexible and gradient speech categorization. On the other hand, less diverse linguistic input led to more categorical speech categorization. We argue that these findings have implications for speech perception as well as linguistic diversity research.

Revitalizing faith: an inquiry into political Sufism and religious continuity in contemporary Kazakhstan.

The article embarks upon a study of political Sufism in contemporary Kazakhstan and patterns of religious continuity through an examination of Islamic manifestations that stretch back centuries, juxtaposed with the current state religious policy. It examines the role of the Muftiate's "official Islam" in shaping the life of religious communities, exploring the intricate interplay between religious identity, secularism, and public perception in the post-Soviet landscape. The author employed a multi-faceted research approach, integrating historical analysis, policy examination, and an ethnographic study of contemporary religious dynamics in the country. The historical analysis provides a foundation of the religious landscape, while policy analysis examines the state's contemporary role in shaping religious practices. Ethnography, focusing on groups such as the Jahriyya, Naqshbandiyya, and the Suhba, involves respondents from diverse regions of Kazakhstan, offering firsthand insights into the experiences and perceptions of Sufi communities at the grassroots level, which enriches the study with a contextual perspective. Notably, fieldwork surveys, where the author engaged with respondents, provide valuable insights into the diverse experiences shaping the dynamics of political Sufism and religious continuity. They involved a diverse group of respondents, including religious leaders, community members, and secular intellectuals, from multiple regions of Kazakhstan at both urban and rural levels. The article is of scientific and practical significance as it enhances understanding of religious dynamics in the country, offering policymakers, scholars, and practitioners valuable perspectives for informed decision-making and policy development. The research's limitations include its focus on specific Sufi groups within Kazakhstan, which may not represent the full spectrum of Islamic practices across the region. The recommendations emphasize the need for further analysis of how prioritizing doctrinal adherence over individual freedoms contributes to the erosion of authentic religious institutions and the politicization of Islam. It also recommends to investigate the roles of Sufi groups in filling spiritual and social vacuums, their potential for politicization, and how this interplay affects religious freedom and national identity in Kazakhstan. This is particularly significant given the state's efforts to appropriate Islamic traditions for ideological purposes, which has led to a separation between Islamic doctrine and its lived expressions. Finally, it emphasizes the need for further comparative research on Sufi movements across Central Asia to better understand how different state policies impact religious communities.

A Unified Multi-Task Learning Model with Joint Reverse Optimization for Simultaneous Skin Lesion Segmentation and Diagnosis.

Classifying and segmenting skin cancer represent pivotal objectives for automated diagnostic systems that utilize dermoscopy images. However, these tasks present significant challenges due to the diverse shape variations of skin lesions and the inherently fuzzy nature of dermoscopy images, including low contrast and the presence of artifacts. Given the robust correlation between the classification of skin lesions and their segmentation, we propose that employing a combined learning method holds the promise of considerably enhancing the performance of both tasks. In this paper, we present a unified multi-task learning strategy that concurrently classifies abnormalities of skin lesions and allows for the joint segmentation of lesion boundaries. This approach integrates an optimization technique known as joint reverse learning, which fosters mutual enhancement through extracting shared features and limiting task dominance across the two tasks. The effectiveness of the proposed method was assessed using two publicly available datasets, ISIC 2016 and PH2, which included melanoma and benign skin cancers. In contrast to the single-task learning strategy, which solely focuses on either classification or segmentation, the experimental findings demonstrated that the proposed network improves the diagnostic capability of skin tumor screening and analysis. The proposed method achieves a significant segmentation performance on skin lesion boundaries, with Dice Similarity Coefficients (DSC) of 89.48% and 88.81% on the ISIC 2016 and PH2 datasets, respectively. Additionally, our multi-task learning approach enhances classification, increasing the F1 score from 78.26% (baseline ResNet50) to 82.07% on ISIC 2016 and from 82.38% to 85.50% on PH2. This work showcases its potential applicability across varied clinical scenarios.

Geometry-Aware Retargeting for Two-Skinned Characters Interaction

Interactive motion between multiple characters is widely utilized in games and movies. However, the method for generating interactive motions considering the character's diverse mesh shape has yet to be studied. We propose a Spatio Cooperative Transformer (SCT) to retarget the interacting motions of two characters having arbitrary mesh connectivity. SCT predicts the residual of root position and joint rotations considering the shape difference between the source and target of interacting characters. In addition, we introduce an anchor loss function for SCT to maintain the geometric distance between the interacting characters when they are retargeted. We also propose a motion augmentation method with deformation-based adaptation to prepare a source-target paired dataset with an identical mesh connectivity for training. In experiments, our method achieved higher accuracy for semantic preservation and produced less artifacts of inter-penetration between the interacting characters for unseen characters and motions than the baselines. Moreover, we conducted a user evaluation using characters with various shapes, spanning low-to-high interaction levels to prove better semantic preservation of our method compared to previous studies.