R1-Track: Direct Application of MLLMs to Visual Object Tracking via Reinforcement Learning
This work addresses the need for more flexible and efficient visual tracking methods, though it appears incremental as it adapts existing MLLMs rather than introducing a fundamentally new approach.
The paper tackled the problem of visual single object tracking by directly applying a multi-modal large language model (MLLM) to overcome limitations of traditional methods, achieving notable performance on the GOT-10k benchmark through fine-tuning with reinforcement learning.
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.