SCVCD-5K: The First Large Scale Screen Content Video Dataset for Compression
IEEE Signal Processing Letters · 2026 · 33 · 3287-3291
DOI: 10.1109/lsp.2026.3722151 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2026. Verfasst von Chaofei Li und 4 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Screen content videos (SCVs) play a vital role in applications such as remote work, online education, and cloud gaming. Due to the huge data volume of SCVs, specific compression technologies are urgent. In recent years, deep learning-based video compression methods that highly depends on training datasets have shown significant performance gain compared with traditional hybrid coding frameworks. However, existing public datasets predominantly focus on natural scenes and human videos, overlooking the unique characteristics of SCVs. To address this gap and advance related research, we construct the first large-scale, multi-type SCV dataset, namely SCVCD-5 K. This dataset encompasses diverse scenarios such as operating system interfaces, office software, and web browsing, comprehensively reflecting the diversity and complexity of screen content. To validate its practicality, extensive experiments were conducted using traditional video codecs and deep learning-based video compression methods. Results demonstrate that the proposed dataset not only provides a standardized benchmark for evaluating the compression performance of SCV but also lays a robust data foundation for designing learning-based compression methods for SCVs. The dataset could be downloaded at https://github.com/yuanhui0325/SCVCD-5~K/blob/main/README.md.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Li, C., Tian, J., Jiang, S., Yang, J., & Yuan, H. (2026). SCVCD-5K: The First Large Scale Screen Content Video Dataset for Compression. IEEE Signal Processing Letters, 33, 3287-3291. https://doi.org/10.1109/lsp.2026.3722151
Vancouver
Li C, Tian J, Jiang S, Yang J, Yuan H. SCVCD-5K: The First Large Scale Screen Content Video Dataset for Compression. IEEE Signal Processing Letters. 2026;33:3287-3291. doi:10.1109/lsp.2026.3722151
BibTeX
@article{Li2026,
author = {Li, Chaofei and Tian, Jinglan and Jiang, Shiqi and Yang, Jingwen and Yuan, Hui},
title = {SCVCD-5K: The First Large Scale Screen Content Video Dataset for Compression},
journal = {IEEE Signal Processing Letters},
year = {2026},
volume = {33},
pages = {3287--3291},
doi = {10.1109/lsp.2026.3722151}
}
EUVIMED