32584 research outputs found
Sort by
Using Facial Recognition for Selective Pose Detection
Get PDFPose detection involves locating and identifying key body points for all individuals within a frame. This enables the ability to convert the pose into a digital format, which can then be recorded and analyzed for a variety of purposes. Advancements in the field have already opened applications in areas such as digital fitness coaches, fall detection, and virtual reality. Existing approaches primarily focus on tracking all detected individuals, which limits the practical applications when attempting to analyze a single or specific subject when there are other people in frame. Previous work has discussed integrating identification, but these approaches use identification primarily to improve tracking consistency rather than to enable targeted detection. This work addresses the understudied problem of targeted pose detection, where only specific individuals are recognized and tracked. A novel multi-task model and pipeline is proposed, integrating techniques for person detection, pose estimation, facial recognition, and object tracking. In addition, a proof of concept implementation is created. The approach enables pose detection for broader use in fields like sports data analytics where the frame includes crowds or other individuals, as well as in security applications such as prisons and airports. By streamlining detection for targeted results, the model provides more useful output while removing unnecessary computation for irrelevant subjects
Quantization on Graph Neural Networks for Image Classification
No full textQuantization has become a key approach for reducing storage and computational demands of deep neural networks while maintaining high accuracy. Although 8-bit quantization is well-established for convolutional architectures such as ResNet50 and MobileNetV2, its application to graph-based vision models remains underexplored. In this work, we extend quantization-aware training to Vision Graph Neural Networks (ViGs) and conduct comparisons with quantized CNNs on the CIFAR-100 dataset. To ensure parity, all models have same training hyperparameters such as learning rate, batch size, optimizer, number of epochs. We used numerous techniques to preserve performance for low-bit precision. First, Pauta Quantization clips activation outliers based on statistical thresholds. Next, Attention Quantization Distillation (AQD) encourages the quantized network to mimic channel-wise attention patterns from full-precision activations to retain feature representations. Finally, Stochastic Quantization Distillation (SQD) injects randomness into the quantization process to make it more robust. Comprehensive experiments on CIFAR-100 reveal that 8-bit quantization delivers an excellent trade-off between efficiency and accuracy across both architectures. Notably, our quantized ViGs almost matches the performance of their ResNet50 and MobileNetV2 for top-1 accuracy, while also making similar inference and memory footprint. These findings demonstrate that quantized Vision GNNs are a practical alternative to CNNs for deployment on resource-constrained devices. Future research will explore fine-tuning of AQD and other distillation parameters
Understanding the State of Open Access Publishing in India: Where Do We Stand? [Cancelled]
No full textThe seed of Open Access publishing was sown at the meeting held in Budapest in 2002, which marked a revolutionary step in scholarly communication. This bold declaration laid the foundation for a movement committed to making research freely and openly available. It will be the silver jubilee of the Open Access (OA) movement next year. In these years, numerous initiatives have emerged to strengthen and expand the reach of Open Access. The OA landscape has evolved by introducing varied publishing models, policy mandates, and platforms. Therefore, this study was conducted among a diverse sample (n=287) of Indian researchers who participated in nationwide capacity-building programmes supported by the Open Research Funders Group (ORFG). The survey aimed to capture insights into how researchers engage with and understand Open Access publishing. The respondents represented a broad spectrum of academic disciplines, including law, journalism, agriculture, chemical sciences, biological sciences, physical sciences, medical and health sciences, social sciences, arts and humanities, engineering, and technology. This disciplinary diversity gave a more holistic understanding of the varying awareness, motivation, and barriers associated with Open Access publishing across different knowledge domains. The study explored various aspects of publishing practices, including researchers’ experience with Open Access journals, key criteria for journal selection, awareness of OA publishing models, motivations for OA publishing, its perceived role in advancing scholarship, encountered challenges, and views on institutional support, policy advocacy, and the need for training and education
Spartan Daily, September 18, 2025
Get PDFVolume 165, Issue 12https://scholarworks.sjsu.edu/spartan_daily_2025/1054/thumbnail.jp
Seasonality in terminus ablation rates for the glaciers in Greenland (Kalaallit Nunaat)
Get PDFMarine-terminating glaciers of Greenland (Kalaallit Nunaat) have been losing mass since the 1990s, with a substantial portion caused by the effects of dynamic change. Conventional assessments of dynamic mass loss, however, often ignore the influence of terminus advance or retreat on the timing of mass loss. Here, we construct and analyze a decade (2013-2023) of monthly ice flux driven both by temporal variability in ice flow (i.e., discharge) and terminus position change - collectively called terminus ablation - for 49 marine-terminating glaciers in Greenland. We calculate terminus ablation rates using open-source datasets, including terminus position, ice surface elevation, ice thickness, and glacier speed. For the majority of glaciers, we observe coincident seasonal variations in terminus position and discharge. However, seasonal variations in terminus ablation are much larger than those in discharge. For the northwest and central west sectors, where the highest fractions of outlet glaciers are included in our terminus ablation dataset, terminus ablation varies by ∼ 51 and ∼ 25 Gt yr−1, respectively, over each year. In contrast, the corresponding variation in discharge is only ∼ 5 Gt yr−1. While our terminus ablation time series do not include every outlet glacier, they suggest that terminus position change is the dominant contributor to Greenland glacier dynamic mass loss at seasonal timescales, in contrast with the relatively small influence of terminus change on decadal-scale mass loss. Since seasonality in mass loss can influence the fate of freshwater fluxes, we recommend that studies concerned with the impacts of Greenland mass loss on local to global ocean properties should account for seasonal terminus position change
Comment on brownian motion of droplets induced by thermal noise
Get PDFWe simulate phase separated fluids using the Cahn-Hillard fluctuating hydrodynamic (CH-FHD) model and measure the statistical properties of capillary waves generated by thermal fluctuations. Our measurements are in good agreement with stochastic lubrication theory and molecular dynamics simulations but differ significantly from recent CH-FHD results by Zhang et al. [Phys. Rev. E 109, 024208 (2024)2470-004510.1103/PhysRevE.109.024208]. Specifically, we find that capillary wave statistics at thermodynamic equilibrium are independent of transport properties, namely viscosity and species diffusion
Optimizing Epoxy Nanocomposites with Oxidized Graphene Quantum Dots for Superior Mechanical Performance: A Molecular Dynamics Approach
Get PDFDue to their excellent mechanical properties, epoxy composites are widely used in low-density applications. However, the brittle epoxy matrix often serves as the principal failure point. Matrix enhancements can be achieved by optimizing polymer combinations to maximize intermolecular interactions or by introducing fillers. While nanofillers such as clay, rubber, carbon nanotubes, and nanoplatelets enhance mechanical properties, they can lead to issues like agglomeration, voids, and poor load transfer. Quantum dots, being the smallest nanofillers, offer higher dispersion and the potential to promote intermolecular interactions, enhancing stiffness, strength, and toughness simultaneously. This study employed molecular dynamics simulations to design graphene quantum dot (GQD) reinforced epoxy nanocomposites. By functionalizing GQDs with oxygen-based groups─hydroxyl, epoxide, carboxyl, and mixed chemistries─their effects on the mechanical properties of nanocomposites were systematically evaluated. Results show that hydroxyl-functionalized GQDs provide optimal performance, increasing stiffness and yield strength by 18.4 and 56.1%, respectively. Structural analysis reveals that these GQDs promote a closely packed molecular configuration, resulting in reduced free volume
A comparison of fit, heat stress, oxygen saturation and comfort between a novel reusable mask and disposable N95 respirator
No full textThe effectiveness of face masks in infection prevention depends not only on filtration technology but also on user compliance. However, existing masks suffer from limitations impacting comfort, ease of use, and communication, leading to reduced compliance, especially during prolonged use in healthcare settings. Innovations in mask design are needed to address these issues to ensure effective protection. To provide insights into novel face mask design aimed at enhancing infection prevention in healthcare settings and to introduce new evaluation methods for novel face masks, a fit and usability study was conducted with 22 volunteers, comparing a novel reusable mask, Altus Hero 1 (Hero), to N95 respirators. Subjects performed Occupational Safety and Health Administration (OSHA)-accepted quantitative fit test and usability test, and completed a post-test survey. The survey assessed communication, breathability, humidity retention, eyeglasses fitting, and long-term wear preference. Face temperature and blood oxygen levels were recorded during testing. Hero showed significantly reduced heat retention (p\u3c0.05) compared to N95, aligning with survey responses indicating Hero felt cooler. No significant differences were found in blood oxygen levels between masks. Despite needing design refinements, most subjects preferred Hero for comfort and usability. This study discusses enhancements in design, fit, comfort, and materials to better meet users’ needs and ensure compliance. It highlights critical and universal design considerations for future face masks and introduces methodological innovations for evaluating mask fit and usability. The findings offer valuable insights for advancing personal protective equipment for preventing infections and future pandemics