Fake-HR1: Rethinking Reasoning of Vision Language Model for Synthetic Image Detection

TL;DR

Fake-HR1 is a hybrid-reasoning model that adaptively determines when to reason, improving synthetic image detection efficiency and accuracy over existing large language models.

Contribution

It introduces a novel adaptive reasoning framework with a two-stage training process for better synthetic image detection.

Findings

Outperforms existing LLMs in reasoning and detection accuracy.

Reduces resource overhead by adaptively skipping unnecessary reasoning.

Demonstrates improved response efficiency in synthetic image detection.

Abstract

Recent studies have demonstrated that incorporating Chain-of-Thought (CoT) reasoning into the detection process can enhance a model's ability to detect synthetic images. However, excessively lengthy reasoning incurs substantial resource overhead, including token consumption and latency, which is particularly redundant when handling obviously generated forgeries. To address this issue, we propose Fake-HR1, a large-scale hybrid-reasoning model that, to the best of our knowledge, is the first to adaptively determine whether reasoning is necessary based on the characteristics of the generative detection task. To achieve this, we design a two-stage training framework: we first perform Hybrid Fine-Tuning (HFT) for cold-start initialization, followed by online reinforcement learning with Hybrid-Reasoning Grouped Policy Optimization (HGRPO) to implicitly learn when to select an appropriate reasoning mode. Experimental results show that Fake-HR1 adaptively performs reasoning across different types of queries, surpassing existing LLMs in both reasoning ability and generative detection performance, while significantly improving response efficiency.