ABSTRACT Malignant lung nodules can significantly affect patients' normal lives and, in severe cases, threaten their survival. Owing to the heterogeneity of computed tomography scans and the varying sizes of nodules, physicians often face challenges in diagnosing this condition. Therefore, a novel adaptive multi‐channel fusion network (AMCF‐Net) is proposed for computer‐aided diagnosis of lung nodules. First, a Multi‐Channel Fusion Model module is designed, which divides the channels into two parts in specific proportions, effectively extracting multi‐scale channel information while reducing network parameters. After the feature maps output at each layer of the AMCF‐Net, a novel adaptive depth‐wise separable convolution with a squeeze‐and‐excitation module is designed to adaptively integrate the feature maps of various stages of the AMCF‐Net, ensuring that the key lesions of lung nodules are not lost during classification. Finally, a hybrid loss scheme based on an adaptive mixing ratio is proposed to solve the problem of an imbalanced number of positive and negative nodule samples in the dataset. The model achieved the following test results: an accuracy of 90.22%, a specificity of 98.19%, an F1‐score of 86.57%, a sensitivity of 86.49%, and a G‐mean of 87.72%. Compared with other advanced networks, AMCF‐net delivers high‐precision lung nodule classification with minimal inference cost. Related codes have been released at: https://github.com/GuYuIMUST/AMCF‐net .

Currently, one of the most well-liked study fields in computer-vision is skeleton-based human action recognition. As the Lie group in Riemannian manifolds is able to precisely describe 3D geometric relationships among rigid bodies, it is widely used in skeleton-based action recognition approaches to construct action feature descriptors. Regrettably, the majority of these approaches overlook crucial body parts and skeletons in favor of focusing solely on spatio-temporal descriptors of an action as a whole. A manifold-based rigid body motion attention mechanism is proposed to assign varying degrees of importance to the relative geometries of different limb motions, and then a skeleton-based spatial attention module is constructed on the basis of limb motions for the more efficient extraction of spatial features from skeletons. Furthermore, a Lie group-based temporal attention mechanism is proposed on the basis of the first two phases to choose significant skeleton frames in an effort to raise the level of action recognition accuracy even higher. Results from experiments on four of the most influential action datasets demonstrate that the proposed approach performs better in terms of action recognition accuracy than many cutting-edge approaches based on skeletons.