SGCn
SGCn, commonly abbreviated as SGCN, stands for Spatial Graph Convolutional Network. It is a class of neural networks designed to operate on graph-structured data by applying convolution-like operations over a graph’s nodes and their neighborhoods. In practice, SGCNs are frequently used with skeleton-based data, where nodes represent body joints and edges represent limbs, to capture the spatial relationships between joints.
The core idea of an SGCN is to represent input signals on a graph using an adjacency
To model evolution over time, SGCNs are often extended to spatio-temporal graphs, combining graph convolutions with
Applications of SGCNs include action recognition from video using sequences of pose data, gesture recognition, and
Historically, the approach gained prominence through skeleton-based action recognition models that apply graph convolutions to human