---
license: cc-by-nc-4.0
task_categories:
- object-detection
language:
- en
pretty_name: LTDv2
---

# LTDv2: A Large-Scale Long-term Thermal Drift Dataset for Robust Multi-Object Detection in Surveillance

The LTDv2 dataset is designed as a single-scene thermal multi-object detection dataset of over 1 million frames extracted from video clips spanning a period of 8 months. The dataset contains more than 6.8 million annoted bounding boxes of 4 object classes (Person, Bicycle, Motorcycle, Vehicle). Encompasing several seasons, weather conditions and day-night cycles, objects and background exhibits drastic variation in appearance over time. Each image is also paired with meta-data detailing weather related environmental conditions in the area, providing supplementary information to aid the visual domain.

Additional details can be found in the [dataset paper](https://www.techrxiv.org/users/942755/articles/1313175-ltdv2-a-large-scale-long-term-thermal-drift-dataset-for-robust-multi-object-detection-in-surveillance).

If you use the LTDv2 dataset, please cite:
```
@article{LTDv2_dataset,
    title={LTDv2: A Large-Scale Long-term Thermal Drift Dataset for Robust Multi-Object Detection in Surveillance},
    DOI={10.36227/techrxiv.175339329.95323969/v1},
    publisher={Institute of Electrical and Electronics Engineers (IEEE)},
    author={Parola, Marco and Aakerberg, Andreas and Johansen, Anders S and Nikolov, Ivan A and Cimino, Mario GCA and Nasrollahi, Kamal and Moeslund, Thomas B},
    year={2025},
}
```