Probing Fine-Grained Action Understanding and Cross-View Generalization of Foundation Models
This work addresses the challenge of cross-view generalization for fine-grained action recognition, which is crucial for domain-specific applications like industrial assembly, but it is incremental as it systematically evaluates existing models and design choices.
The paper tackled the problem of how perspective changes affect foundation models in fine-grained human activity recognition, finding that high accuracies on standard benchmarks can be misleading due to overlooked real-world factors like varying camera views.
Foundation models (FMs) are large neural networks trained on broad datasets, excelling in downstream tasks with minimal fine-tuning. Human activity recognition in video has advanced with FMs, driven by competition among different architectures. However, high accuracies on standard benchmarks can draw an artificially rosy picture, as they often overlook real-world factors like changing camera perspectives. Popular benchmarks, mostly from YouTube or movies, offer diverse views but only coarse actions, which are insufficient for use-cases needing fine-grained, domain-specific actions. Domain-specific datasets (e.g., for industrial assembly) typically use data from limited static perspectives. This paper empirically evaluates how perspective changes affect different FMs in fine-grained human activity recognition. We compare multiple backbone architectures and design choices, including image- and video- based models, and various strategies for temporal information fusion, including commonly used score averaging and more novel attention-based temporal aggregation mechanisms. This is the first systematic study of different foundation models and specific design choices for human activity recognition from unknown views, conducted with the goal to provide guidance for backbone- and temporal- fusion scheme selection. Code and models will be made publicly available to the community.