HDO is a high-quality infrared and visible video sequence dataset. The dataset consists of various HD video sequences captured under different scenes, aimed to facilitate research and development in the fields of image processing and computer vision.

Dataset Description

The HDO dataset contains a large number of high-definition video sequences. The main scenes captured in the HDO dataset are streets and campuses. We sampled each video at 100ms intervals to obtain video sequences.

The features of the original sampled video sequence include:

• Data format: JPG
• Image resolution: 1280 x 1024 (infrared) ,1920 x 1080 (visible)
• Dataset size: approximately 8.5 GB

To facilitate applications on vision tasks such as HDO dataset fusion, registration, and tracking, we also aligned most of them to form 5788 pairs of paired infrared and visible data.

The features of the paired video sequence include:

• Data format: JPG
• Image resolution: 640 x 480 (infrared) ,640 x 480 (visible)
• Dataset size: approximately 1.0 GB

Dataset Preview

Download

Raw video sequences

• Google Drive
• Baidu Yun

Aligned image pairs

• Google Drive
• Baidu Yun

Note that the dataset is for academic and non-commercial use only. If this dataset was helpful to you, please feel free to cite its source. Any questions or issues accessing the dataset, please contact us at xiehoushengsc@163.com.

---

RCVS: Round-the-Clock Video Stream registration and fusion scheme