A Collection of 100 4K Video Sequences
University of Bristol
University of Bristol
University of Bristol
University of Bristol
About
This is a collection of 100 openly available UHD video sequences. The sequences were downloaded from a variety of sources, more specifically, Netflix Chimera, Ultra Video Group, Harmonic Inc, SJTU, and AWS Elemental. All the sequences were spatially cropped to UHD (if originally 4K resolution, 4096 × 2160 pixels), converted to 4:2:0 chroma subsampling (if originally 4:2:2 or 4:4:4) and temporally cropped to 64 frames. We used 64 frames to reduce the time required to encode all the training sequences. In addition, each sequence contains a single scene (without scene cuts) and the majority of the test sequences have a frame rate of 60 fps and bit depth of 10 bits per sample. Further detailed specifications of all training sequences can be found in the tables in the next pages. Please check the pdf below for the full list of specifications of the sequences and their origin.
Thumbnails
Downloads
- Report on Dataset Specifications [pdf]
- Fill in this Form to get access to the dataset.
- Papers that used this dataset
- A. V. Katsenou, J. Sole, and D. R. Bull, "Efficient Bitrate Ladder Construction for Content-Optimized Adaptive Video Streaming", IEEE Open Journal of Signal Processing, 2021 (early access). [Paper]
- A. V. Katsenou, J. Sole, and D. R. Bull, "Content-gnostic Bitrate Ladder Prediction for Adaptive Streaming," 2019 Picture Coding Symposium. [Paper]