R1-Track: Direct Application of MLLMs to Visual Object Tracking via Reinforcement Learning

TL;DR

This paper introduces R1-Track, a novel approach that applies multi-modal large language models directly to visual object tracking using reinforcement learning, enabling flexible initialization and competitive performance.

Contribution

We fine-tuned a large language model with reinforcement learning for visual tracking, achieving flexible initialization and strong results on benchmark datasets.

Findings

R1-Track performs well on GOT-10k benchmark.

Supports initialization via bounding boxes or text descriptions.

Retains most of the original model's general capabilities.

Abstract

Visual single object tracking aims to continuously localize and estimate the scale of a target in subsequent video frames, given only its initial state in the first frame. This task has traditionally been framed as a template matching problem, evolving through major phases including correlation filters, two-stream networks, and one-stream networks with significant progress achieved. However, these methods typically require explicit classification and regression modeling, depend on supervised training with large-scale datasets, and are limited to the single task of tracking, lacking flexibility. In recent years, multi-modal large language models (MLLMs) have advanced rapidly. Open-source models like Qwen2.5-VL, a flagship MLLMs with strong foundational capabilities, demonstrate excellent performance in grounding tasks. This has spurred interest in applying such models directly to visual tracking. However, experiments reveal that Qwen2.5-VL struggles with template matching between image pairs (i.e., tracking tasks). Inspired by deepseek-R1, we fine-tuned Qwen2.5-VL using the group relative policy optimization (GRPO) reinforcement learning method on a small-scale dataset with a rule-based reward function. The resulting model, R1-Track, achieved notable performance on the GOT-10k benchmark. R1-Track supports flexible initialization via bounding boxes or text descriptions while retaining most of the original model's general capabilities. And we further discuss potential improvements for R1-Track. This rough technical report summarizes our findings as of May 2025.