RAVEN: Robust Advertisement Video Violation Temporal Grounding via Reinforcement Reasoning
This addresses the problem of ensuring platform compliance for advertisement video violation detection, with incremental improvements in temporal grounding and generalization.
The paper tackles the problem of precise temporal grounding for advertisement video violation detection, which struggles with noisy annotations and limited generalization, by proposing RAVEN, a framework that integrates curriculum reinforcement learning with multimodal large language models; experiments show it achieves superior performance in violation category accuracy and temporal interval localization on industrial datasets and public benchmarks, with online A/B testing validating significant improvements in precision and recall.
Advertisement (Ad) video violation detection is critical for ensuring platform compliance, but existing methods struggle with precise temporal grounding, noisy annotations, and limited generalization. We propose RAVEN, a novel framework that integrates curriculum reinforcement learning with multimodal large language models (MLLMs) to enhance reasoning and cognitive capabilities for violation detection. RAVEN employs a progressive training strategy, combining precisely and coarsely annotated data, and leverages Group Relative Policy Optimization (GRPO) to develop emergent reasoning abilities without explicit reasoning annotations. Multiple hierarchical sophisticated reward mechanism ensures precise temporal grounding and consistent category prediction. Experiments on industrial datasets and public benchmarks show that RAVEN achieves superior performances in violation category accuracy and temporal interval localization. We also design a pipeline to deploy the RAVEN on the online Ad services, and online A/B testing further validates its practical applicability, with significant improvements in precision and recall. RAVEN also demonstrates strong generalization, mitigating the catastrophic forgetting issue associated with supervised fine-tuning.