Overcoming Topology Agnosticism: Enhancing Skeleton-Based Action Recognition through Redefined Skeletal Topology Awareness

arXiv:2305.11468v3 Announce Type: replace
Abstract: Graph Convolutional Networks (GCNs) have long defined the state-of-the-art in skeleton-based action recognition, leveraging their ability to unravel the complex dynamics of human joint topology through the graph's adjacency matrix. However, an inherent flaw has come to light in these cutting-edge models: they tend to optimize the adjacency matrix jointly with the model weights. This process, while seemingly efficient, causes a gradual decay of bone connectivity data, culminating in a model indifferent to the very …

abstract action recognition art arxiv cs.cv dynamics edge graph human light matrix networks recognition state through topology type