Title: 
BVI-DVC: A Training Database for Deep Video Compression

Description: 
This .zip file contains 800 sequences within the BVI-DVC training database. These are selected from
1) Videvo Free Stock Video Footage set (Stock footage provided by Videvo, downloaded from https://www.videvo.net)
2) IRIS32 Free 4K Footage set (https://www.youtube.com/channel/UCjJee-JAzoRRH5T0wqe7_tw)
3) Harmonics database (https://www.harmonicinc.com/)
4) BVI-Texture database (https://data.bris.ac.uk/data/dataset/1if54ya4xpph81fbo1gkpk5kk4)
5) MCML 4K video quality database (http://mcml.yonsei.ac.kr/downloads/4kuhdvideoquality)
6) BVI-HFR database (https://data.bris.ac.uk/data/dataset/ca830349ffcba535c2045ca9d2304faf)
7) SJTU 4K video database (http://medialab.sjtu.edu.cn/web4k/index.html)
8) LIVE-Netflix database (https://live.ece.utexas.edu/research/LIVE_NFLXStudy/nflx_index.html)
9) Mitch Martinez Free 4K Stock Footage set (https://mitchmartinez.com/free-4k-red-epic-stock-footage/)
10) Dareful Free 4K Stock Video data set (https://www.dareful.com/)
11) MCL-V database (http://mcl.usc.edu/mcl-v-database/)
12) MCL-JCV database (http://mcl.usc.edu/mcl-jcv-dataset/)
13) Netflix Chimera (https://www.netflix.com/gb/title/80015538)
14) TUM HD databases (https://www.ei.tum.de/en/ldv/research/videolab/data-sets-downloads/tum-1080p50-data-set/)
15) Ultra Video Group-Tampere University database (http://ultravideo.cs.tut.fi/#testsequences)

All of these were truncated from their original length to 64 frames without any scene cuts. The sequences in this database was carefully selected to cover a wide range of video scenes and texture types that will be beneficial to improvement of generalisation ability of CNN models.

More information could be found from the dedicated webpage: 
https://vilab.blogs.bristol.ac.uk/?p=2375

If this content has been mentioned in a research publication, please give credit to the University of Bristol, by referencing the following paper:
[1] Di Ma, Fan Zhang, and David, R. Bull, "BVI-DVC: A Training Database for Deep Video Compression", arXiv preprint arXiv:2003.13552 (2020)

How to use BVI-DVC database?

1. Before using the database, all .mp4 files must be decompressed as 4:2:0 YCbCr files. You can:
		1) Unzip the BVI-DVC.zip file.
		2) Download ffmpeg software from https://www.ffmpeg.org/download.html, and place the "ffmpeg-hi10-heaac.exe" file alongside the "decompress.m" file.
		3) Using MATLAB to run the "decompress.m" file.

2. The 800 videos can be used to train CNN models for deep video compression or other computer video tasks, such as image/video super-resolution, image/video enhancement, video frame interpolation, etc..

If there is any issue with the use of the database, please send emails to:

di.ma@bristol.ac.uk
fan.zhang@bristol.ac.uk
dave.bull@bristol.ac.uk


Dataset Name:
BVI-DVC

Dataset Seq. Num:
800

Copyright disclaimer: 

This database has been compiled by the University of Bristol, Bristol, UK, comprising sequences originally generated by various sources. All intellectual property rights remain with the originators of each sequence. The test sequences from sources (9) and (15) mentioned above shall only be used for academic research (no commercial use). Material from other sources can also be employed for developing future video coding standards and for evaluating performance of test models in JVET and the subsequent standardization project, and for the relational parent body activities. This copyright and permission notice shall be duplicated whenever the data is copied. The University of Bristol makes no warranties with respect to the material and expressly disclaims any warranties regarding its fitness for any purpose. Unless the above conditions are agreed to by the recipient, no permission is granted for any use and copying of the data. By using the database and sequences, the user agrees to the conditions of this copyright and disclaimer.

Contributer: University of Bristol

Video Properties:
Resolutions and scanning format: From 270p to 2160p and progressive scanned.
Chroma Sampling: 4:2:0
Bit Depth: 10 bit
Content type: Static, Dynamic and Mixed Textures
Special Attributes: N/A
Audio Type: No Audio
File Size: 396G
Run Time: N/A