Join Meta AI's Video Similarity Challenge And Win Up To $50K
Meta AI has announced the Video Similarity Challenge with $100k in total winnings. This competitions comes as a sequel to last year's Image Similarity Challenge, both challenges motivated by the spread of harmful content on social media.
Created on December 12|Last edited on December 12
Comment
Meta AI, in collaboration with Driven Data, has announced the Video Similarity Challenge. This competition asks entrants to build an AI model which can tell if a video has been manipulated in some way. Up to $50k in winnings is at stake, so if your ML team has any good ideas, be sure to apply.
Like the Image Similarity Challenge hosted around a year ago, the Video Similarity Challenge is motivated by the difficulty of large-scale content moderation on social media platforms. It's no secret that social media is constantly barraged with spam content, so the hope is that AI models can improve and automate moderating this kind of harmful content.
Video Similarity Challenge details
The Video Similarity Challenge features two tracks: Descriptor Track and Matching Track. The former tasks entrants to develop a model which generates useful vector representations of videos, while the latter tasks entrants to develop a model which can match manipulated videos with their original counterparts.
The first phase of the competition runs from December 2nd to March 24th, where entrants have time to develop their models and post their progress on the public leaderboard. The second phase runs from April 2nd to April 9th, where entrants can submit up to three models to test against unseen test data and where competition winners will be selected.
Each track has identical prize pools, the first place getting $25k, the second getting $15k, and third place taking home $10k, for a total of $100k in winnings; If you place first in both, you can get your hands on $50k.
Find out more
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.