TreeAgent: A Generalizable Multi-Agent Framework for Automated Bias Labeling in Forestry via Compiled Expert Rules and Vision-Language Models

2026-06-30

AI Summary

This study addresses the challenges in forestry remote sensing where tree height deviation annotations rely heavily on manual labeling, resulting in low efficiency and poor consistency. To overcome these limitations, the authors propose the Decoupled Declarative Decision (D3) framework, which constructs a multi-agent system that incorporates expert-defined decision trees as structural priors. By integrating vision-language models at the node level for semantic awareness and employing multi-agent voting to mitigate model stochasticity, the approach achieves high interpretability and annotation efficiency without requiring modifications to accommodate different expert rules. Experimental results demonstrate that the proposed method significantly outperforms supervised learning baselines in tree height deviation classification, substantially reducing the annotation burden on domain experts while maintaining strong interpretability.

Abstract

Human-labeled data are widely used as reference annotations in ML, despite known variability across annotators in many expert-driven domains. In addition, expert annotation is slow, inconsistent, and remains a major bottleneck for scaling tasks like tree height bias classification in forestry remote sensing. We propose a multi-agent system (MAS) that orchestrates expert decision trees with Vision-Language Models (VLMs), treating the decision tree as a structural prior while VLMs perform localized semantic perception at individual nodes, with multi-agent voting to mitigate VLM stochasticity. We formalize a Decoupled Declarative Decision (D3) Framework that enables zero-modification generalization across diverse expert-defined decision structures. On a tree bias classification testbed, our framework outperforms supervised ML baselines and reduces the amount of expert labeling effort required. These results suggest that agentic orchestration of VLMs with expert priors can reproduce expert-defined labeling procedures at substantially lower annotation cost while maintaining interpretability.

Problem

• bias labeling
• forestry remote sensing
• expert annotation
• tree height bias
• annotation bottleneck

Innovation

• multi-agent system
• vision-language models
• expert rules
• bias labeling
• zero-modification generalization