Reasoning-Enhanced Few-Shot Human Activity Recognition via Visual Reinforcement Learning
IEEE Access · 2026 · 14 · 70089-70101
DOI: 10.1109/access.2026.3691141 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Wenqi Zheng und 1 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 44 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: JSPS KAKENHI, China Scholarship Council during the doctoral study at Kyushu University.
Abstract (Originalsprache)
Most existing human activity recognition (HAR) models rely on large-scale labeled datasets for training, which makes it difficult for HAR models to recognize activity classes unseen during training when only a few samples are available. Moreover, they struggle to capture contextual information, such as semantic relationships among activities, which further limits their robustness in real-world scenarios. In this work, we introduce FAVOR++, a reinforcement learning (RL)-based framework that enables Multimodal Large Language Models (MLLMs) to perform structured and interpretable reasoning directly over human-activity image inputs and task prompts, thereby improving generalization to novel actions. FAVOR++ first constructs a reasoning-enhanced chain-of-thought dataset by sampling informative plan–reason trajectories, which are used to perform chain-of-thought supervised fine-tuning. Building on this, FAVOR++ applies Group Relative Policy Optimization (GRPO) for post-training, using an outcome-based reward to support classification. To further stabilize training and enhance downstream generalization, FAVOR++ incorporates rectified ratio clipping and dynamic sampling, which suppress unstable updates and reduce variance during policy optimization. Extensive experiments across five few-shot HAR datasets and five distinct few-shot settings demonstrate that FAVOR++ consistently outperforms state-of-the-art methods. Notably, under the 1-shot setting on UCF-101, the proposed method improves the classification accuracy from 38.41% to 72.35%. Furthermore, in the cross-domain generalization scenario, FAVOR++ boosts the average 1-shot accuracy across five in-domain and out-of-domain datasets from 36.18% to 51.99%. In addition to these performance gains, the model also provides more interpretable reasoning and exhibits stronger generalization capability.
Förderung
- JSPS KAKENHI
- China Scholarship Council during the doctoral study at Kyushu University
Zitieren
APA
Zheng, W., & Arakawa, Y. (2026). Reasoning-Enhanced Few-Shot Human Activity Recognition via Visual Reinforcement Learning. IEEE Access, 14, 70089-70101. https://doi.org/10.1109/access.2026.3691141
Vancouver
Zheng W, Arakawa Y. Reasoning-Enhanced Few-Shot Human Activity Recognition via Visual Reinforcement Learning. IEEE Access. 2026;14:70089-70101. doi:10.1109/access.2026.3691141
BibTeX
@article{Zheng2026,
author = {Zheng, Wenqi and Arakawa, Yutaka},
title = {Reasoning-Enhanced Few-Shot Human Activity Recognition via Visual Reinforcement Learning},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {70089--70101},
doi = {10.1109/access.2026.3691141}
}
EUVIMED