3MASSIV
X
X
A large scale multilingual, multiaspect and multimodal dataset of expertly-annotated short videos from Moj.
3MASSIV is a human-annotated, culturally diverse, multimodal and multilingual dataset containing short videos uploaded to short video application - Moj.
34 popular social media concepts
Comedy
Romance
Pranks
Fails
Memes
Food
Dance
Music
Fashion
... and many more !
11 Indic Languages
50K labelled and 100K unlabelled videos short videos (~10-20 seconds)
Diverse Video Types
Reaction Videos
Split Screens
Self-Shot Videos
Animations
Movie/TV-Shows Clips
Diverse Audio Types
Self-sung songs
Lip-syncs
Dialogues
Monologues
Background Music
Human Annotations
Every video has been annotated by three expert reviewers with each language having dedicated reviewers.
3MASSIV is very rich in annotations and can be used for exploring various research directions:
Concept Understanding
Affective Computing
Language Prediction
Media Format Identification
Temporal Analysis
Team
License
3MASSIV dataset is available for only reseach purposes. Any commercial use of the dataset is strictly prohibited.