MAC 2026: Advancing Micro-Action Analysis Towards Fine-Grained Understanding
Quick Answer
This paper shows that The 3rd Micro-Action Analysis Grand Challenge (MAC 2026) aims to enhance fine-grained understanding of micro-actions using multimodal large language models.
Quick Take
This edition expands beyond recognition to evaluate models' abilities in capturing subtle human behaviors, with publicly accessible datasets and protocols established for academic research.
Key Points
- Introduces fine-grained micro-action understanding as a new evaluation task.
- Builds on previous MAC editions with standardized evaluation settings.
- Utilizes multimodal for deeper interpretation.
- Summarizes datasets, evaluation protocols, and competition results.
- Discusses future directions for micro-action analysis in video understanding.
DeepSignal Analysis
What happened
The 3rd Micro-Action Analysis Grand Challenge (MAC 2026) focuses on enhancing the understanding of micro-actions through multimodal large language models. This edition introduces a new task for fine-grained micro-action understanding, building on previous challenges that established standardized evaluation settings and publicly accessible datasets.
Key evidence
- Micro-Actions are defined as subtle human behaviors that provide non-verbal cues, making them challenging to annotate and evaluate due to their short duration and fine-grained semantic differences.
- The first two editions of MAC established standardized evaluation settings for micro-action recognition and detection, contributing to the development of publicly accessible datasets and protocols.
- The 3rd MAC, held in conjunction with ACM Multimedia 2026, introduces a task aimed at assessing models' abilities to interpret subtle human micro-actions using multimodal large language models.
Why it matters
This challenge is significant as it aims to deepen the understanding of micro-actions, which are crucial for human-centric video analysis and affective communication. By expanding the evaluation criteria beyond recognition to include fine-grained understanding, it encourages advancements in model capabilities and applications in social interaction analysis.
Paper Resources
Source Excerpt
Micro-Actions (MAs) are subtle and spontaneous human behaviors that provide important non-verbal cues in social interaction and affective communication. However, their short duration, weak motion patterns, and fine-grained semantic differences make them difficult to annotate, model, and evaluate in a standardized manner. To promote academic research on micro-action analysis, we proposed and have annually organized the Micro-Action Analysis Grand Challenge (MAC) as a public benchmark platform for
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.