AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Code-in-the-Loop Forensics: Agentic Tool Use for Image Forgery Detection

Fanrui Zhang, Qiang Zhang, Sizhuo Zhou, Jianwen Sun, Chuanhao Li, Jiaxin Ai, Yukang Feng, Yujie Zhang, Wenjie Li, Zizhen Li, Yifan Chang, Jiawei Liu and 1 more

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Multimodal LLM benchmarks · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Cold-Start SFT stage before RFT (ablation)ForenAgent · FABenchaccuracy (%): 78.7 -> 89.5Improved
add RFT with GRPO after cold start (ablation)ForenAgent · FABenchaccuracy (%): 81.4 -> 89.5Improved
add tool usage reward (ablation)ForenAgent RFT · FABenchaccuracy (%): 83.8 -> 89.5Improved
add tool logic reward (ablation)ForenAgent RFT · FABenchaccuracy (%): 84.4 -> 89.5Improved
add global forensic priority reward (ablation)ForenAgent RFT · FABenchaccuracy (%): 86.4 -> 89.5Improved
add crop sensitivity reward (ablation)ForenAgent RFT · FABenchaccuracy (%): 87.8 -> 89.5Improved
add reasoning coherence reward (ablation)ForenAgent RFT · FABenchaccuracy (%): 88.0 -> 89.5Improved
use Qwen2.5-VL-7B instead of InternVL3-8B backbone (ablation)ForenAgent · FABenchaccuracy (%): 79.8 -> 89.5Improved
use Qwen2.5-VL-7B instead of LLaVA-7B backbone (ablation)ForenAgent · FABenchaccuracy (%): 75.8 -> 89.5Improved
equip zero-shot MLLM with forensic toolboxGPT-4.1 · FABenchaccuracy (%): 51.4 -> 64.7Improved
equip zero-shot MLLM with forensic toolboxQwen2.5-VL-72B · FABenchaccuracy (%): 50.0 -> 57.4Improved
ForenAgent vs trained Qwen2.5-VL-7B without toolsQwen2.5-VL-7B · SIDA-Testaccuracy (%): 72.9 -> 83.1Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy