OTMJ_TIR

The OTMJ_TIR (Thermal Infrared Dataset for Object Tracking in Mountain Jungle) dataset is an TIR dataset used for object tracking in mountainous jungle scenes.

News

[January 10, 2026]

We have open-source the OTMJ_TIR dataset, which now includes 50 drone target tracking sequences in mountainous and jungle scenes.

Introduction

Mountain jungle environments constitute critical components of Earth's land surface and represent important scenarios for UAV object tracking. However, complex terrain, dense vegetation, and variable weather conditions make object tracking tasks in these environments particularly challenging for drones. The lack of high-quality dedicated datasets has severely constrained the accuracy and robustness of existing infrared object tracking algorithms. To address this gap, this paper introduces the OTMJ_TIR (Object Tracking in Mountain Jungle_TIR) dataset, specifically designed for single-object tracking in mountain jungle environments using infrared imagery. Collected via UAV platforms equipped with infrared sensors, OTMJ_TIR covers jungle scenes under complex climatic and topographic conditions, providing precise target position annotations across 50 video sequences comprising 8.5k high-resolution infrared frames. Notably, the infrared component of OTMJ_TIR is aligned with target positions in our previously released visible-light OTMJ dataset, offering strong support for multi-modal tracking tasks. Experimental results highlight the dataset's challenging nature, providing an important benchmark and resource for advancing infrared remote sensing object tracking technology in extreme environments.

Additionally, OTMJ TIR is fully aligned with our previously released visible-light [OTMJ] dataset in terms of video sequences, flight paths, timestamps, and target bounding box annotations, laying a solid foundation for future RGB-TIR multi-modal fusion tracking and cross-modal target association studies in mountain jungles.

Obtain Dataset

The dataset can be accessed through this link: [Baidu Netdisk], Extract code: 1234.