Home   >  Competitions   > 

 

Background

 

In recent years, deep learning has made great progress in several fundamental tasks such as image recognition and speech recognition, but there are still many problems to be explored in the field of video content understanding.

 

The TikTok (Douyin overseas edition) APP, a product of Bytedance, which is a short video social platform powered by music, has received widely acclaim from users around the world. In the meantime, the understanding and recommendation technology of short video has been the focus of our company's attention.

 

A picture is worth thousands of words. One picture contains a large amount of information. It is difficult to describe it in a few words, not to mention more rich media such as a short video. Facing the problem of understanding the content of short video, Bytedance is a company that has tons of short video clips and the behavior data of hundreds of millions of users. Therefore, while facing the video understanding challenge, we have enough video content and user behavior data to predict users' preference for short video.

 

This challenge provides multi-modal video features, including visual features, text features and audio features, as well as user interactive behavior data, such as click, like, and follow. Each participant needs to model the user's interest through a video and user interaction behavior data set, and then predict the user's click behavior on another video dataset. Both datasets do not overlap in terms of video id.

 

The rank of our challenge accords to the model and predicted results submitted by the participants, based on a predefined score specified in the evaluation criteria.

 

 

Task

 

Predict the probability of each user weather finish watching and like the video on each item of test dataset.

 

We use AUC (area under ROC curve) as our challenge metric. The higher the AUC, the higher the ranking.

 

TRACK1

Massive datasets, hundreds of millions of data information.

 

TRACK2

Small-scale data sets, million-level data information. Baseline method To be available in January 2019.

 


Discussion Board

 

If you have any questions, please ask it in post in the discussion webpage. For questions of dataset, please send an email to AI-Lab-challenge@bytedance.com. For questions of submission, please send an email to support@biendata.com.

Short Video Understanding Challenge

$20,000

56 Teams

start

Final Submissions

2019-01-29

2019-04-22