A fine-grained football action recognition model combining spatio-temporal compression sequences and multi-head cross-attention surpassed existing methods on the SoccerNet-v2 benchmark ...
A neural network-accelerated ice flow model reconstructed Himalayan glacier thickness from satellite velocity data with accuracy comparable to leading global ...
Researchers developed a confidence-gated convolutional neural network and bidirectional recurrent model that enhances rare arrhythmia classification on electrocardiogram signals while ...
A retrospective study in Frontiers in Systems Biology found that ResNet-based AI models achieved high accuracy in classifying breast cancer ...
Researchers have introduced NavLite-AttNet, a lightweight convolutional neural network incorporating hybrid attention mechanisms to classify subtle Navarasa emotional expressions with ...
A lightweight graph-based model that maps relationships between emotion categories achieved high facial expression recognition accuracy on several benchmark datasets ...
A combined DFT and machine learning study maps how the carbon–fluorine bond in tetrafluoromethane breaks across transition metal surfaces, single-atom ...
A case study of the Damiao–Hongshila belt shows geographically weighted models improve mineral prospectivity mapping under spatially non-stationary conditions.
A two-stage diffusion framework with automated assessment generated a gymnasium field-level plan for a Xiong'an campus project that researchers report ...
A multi-scale 3D-CNN tuned by a quantum-inspired metaheuristic optimizer classified eight basketball tactical patterns from player tracking data with 94.3% ...
Embark on a thrilling journey of discovery with Scienmag.com—your ultimate source for cutting-edge breakthroughs. Immerse yourself in a world where curiosity knows no limits and tomorrow’s possibilities become today’s reality!
