Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Frequency-Semantic Enhanced Variational Autoencoder for Zero-Shot Skeleton-based Action Recognition

About

Zero-shot skeleton-based action recognition aims to develop models capable of identifying actions beyond the categories encountered during training. Previous approaches have primarily focused on aligning visual and semantic representations but often overlooked the importance of fine-grained action patterns in the semantic space (e.g., the hand movements in drinking water and brushing teeth). To address these limitations, we propose a Frequency-Semantic Enhanced Variational Autoencoder (FS-VAE) to explore the skeleton semantic representation learning with frequency decomposition. FS-VAE consists of three key components: 1) a frequency-based enhancement module with high- and low-frequency adjustments to enrich the skeletal semantics learning and improve the robustness of zero-shot action recognition; 2) a semantic-based action description with multilevel alignment to capture both local details and global correspondence, effectively bridging the semantic gap and compensating for the inherent loss of information in skeleton sequences; 3) a calibrated cross-alignment loss that enables valid skeleton-text pairs to counterbalance ambiguous ones, mitigating discrepancies and ambiguities in skeleton and text features, thereby ensuring robust alignment. Evaluations on the benchmarks demonstrate the effectiveness of our approach, validating that frequency-enhanced semantic features enable robust differentiation of visually and semantically similar action clusters, improving zero-shot action recognition.

Wenhan Wu, Zhishuai Guo, Chen Chen, Hongfei Xue, Aidong Lu• 2025

Related benchmarks

TaskDatasetResultRank
Action RecognitionNTU RGB+D X-sub 120
Accuracy50.2
430
Action RecognitionNTU RGB-D Cross-Subject 60
Accuracy57.2
336
Action RecognitionNTU-60 (xsub)
Accuracy79.4
223
Action RecognitionNTU-60 48/12 split
Top-1 Acc56.2
103
Action RecognitionNTU-120 96/24 split
Top-1 Acc50.2
84
Action RecognitionNTU RGB+D 120 (110/10 Xsub)
Accuracy72.7
66
Action RecognitionNTU 60 (55/5 split)
Top-1 Acc77
57
Action RecognitionNTU-120 110/10 split--
56
Action RecognitionNTU-RGB+D 60 (48/12)
Accuracy57.2
49
Action RecognitionPKU-MMD 46/5 I (Xsub)
Accuracy71.2
43
Showing 10 of 26 rows

Other info

Follow for update