Skip to content

Datasets related to sports video analysis.

Notifications You must be signed in to change notification settings

abdkhanstd/Sports

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Benchmark datasets for sports video analysis.

About the Dataset

The data set contains a total of 104 full-length sports videos having a cumulative duration of more than 230 hours. The videos were captured in an “as it is form” from internet sources. These videos contain commercials, TV logos, some logos and themes superimposed by bloggers, etc. These videos belong to 12 different categories of sports, that is why they heavily vary in length. Some of the sports categories are time-based sports and some are score based sports such as cricket, baseball etc. All the videos are in mp4 format. Here are some statistics about the videos:

Sports cateagory Number of videos Frame rate range Cummulative durations (HH:MM)
Cricket 11 23∼30 42:23
Rugby 10 23∼30 21:28
Soccer 22 23∼30 39:08
Basketball 11 25∼30 17:59
Baseball 7 25∼30 18:21
Football 8 25∼30 18:51
Tennis 7 27∼30 17:14
Handball 9 24∼30 12:11
Snooker 4 25~30 06:07
Volleyball 4 25∼25 05:48
Ice Hockey 7 25∼30 16:07
Hockey 4 25∼30 08:04

Some selected frame samples

Some samples from the dataset

The complexity of the dataset can clearly be seen from the sample frames. Sports videos are challenging due to the huge amount of cluttering in the background and 3D-induced marketing items which are usually painted on the playfield.

Download Videos

The videos in this dataset are approximate 105 GigaBytes in size with varying qualities. These videos are shared via Microsoft OneDrive business account (other mirrors can be arranged on demand. Please refer to contact info.) Videos can be downloaded from here

Download Converted Audio files

For ease to the users and synchronize the timeline, pre-extracted audio files (.mp3 format) can be downloaded from here

Description of JSON files

The scorebox availability and location data can be found in the “scorebox availability and location” folder. The corresponding JSON data file contains the ground truth “Time”, “Availability” and “Location” of the scorebox. Please note that the time of the video is represented in seconds. The “Location” contains four parameters Ymin, Ymax, Xmin, and Xmax. Ymin represents the starting and Ymax represents the end Y-axis pixel location of ground truth SB. Vice versa, Xmin represents the starting and Xmax represents the end X-axis pixel location of ground truth SB. The starting and end time of an event, with respect to the video, can be found in the representative JSON file in the Events folder.

How to cite?

@article{DBLP:journals/npl/k2020,
  author = {Abdullah Aman Khan and
                        Jie Shao and
                        Waqar Ali and
                        Saifullah Tumrani},
       title = "{Content-Aware Summarization of Broadcast Sports Videos:An Audio–Visual Feature Extraction Approach}",
       journal = {Neural Processing Letters},
       pages = {1--24},
       year = {2020},
}

Contact

Abdullah K. abdkhan@std.uestc.edu.cn Please note that I may respond late but I will.

Important note

More details (code + results) will be added later.

About

Datasets related to sports video analysis.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published