Research

Paper

AI LLM March 09, 2026

Novel Semantic Prompting for Zero-Shot Action Recognition

Authors

Salman Iqbal, Waheed Rehman

Abstract

Zero-shot action recognition relies on transferring knowledge from vision-language models to unseen actions using semantic descriptions. While recent methods focus on temporal modeling or architectural adaptations to handle video data, we argue that semantic prompting alone provides a strong and underexplored signal for zero-shot action understanding. We introduce SP-CLIP, a lightweight framework that augments frozen vision-language models with structured semantic prompts describing actions at multiple levels of abstraction, such as intent, motion, and object interaction. Without modifying the visual encoder or learning additional parameters, SP-CLIP aligns video representations with enriched textual semantics through prompt aggregation and consistency scoring. Experiments across standard benchmarks show that semantic prompting substantially improves zero-shot action recognition, particularly for fine-grained and compositional actions, while preserving the efficiency and generalization of pretrained models.

Metadata

arXiv ID: 2603.08289
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-09
Fetched: 2026-03-10 05:43

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.08289v1</id>\n    <title>Novel Semantic Prompting for Zero-Shot Action Recognition</title>\n    <updated>2026-03-09T12:07:55Z</updated>\n    <link href='https://arxiv.org/abs/2603.08289v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.08289v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Zero-shot action recognition relies on transferring knowledge from vision-language models to unseen actions using semantic descriptions. While recent methods focus on temporal modeling or architectural adaptations to handle video data, we argue that semantic prompting alone provides a strong and underexplored signal for zero-shot action understanding. We introduce SP-CLIP, a lightweight framework that augments frozen vision-language models with structured semantic prompts describing actions at multiple levels of abstraction, such as intent, motion, and object interaction. Without modifying the visual encoder or learning additional parameters, SP-CLIP aligns video representations with enriched textual semantics through prompt aggregation and consistency scoring. Experiments across standard benchmarks show that semantic prompting substantially improves zero-shot action recognition, particularly for fine-grained and compositional actions, while preserving the efficiency and generalization of pretrained models.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-09T12:07:55Z</published>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Salman Iqbal</name>\n    </author>\n    <author>\n      <name>Waheed Rehman</name>\n    </author>\n  </entry>"
}