metadata

language:
  - en
license: cc
size_categories:
  - 10K<n<100K
task_categories:
  - video-text-to-text
  - video-classification
dataset_info:
  features:
    - name: mp4
      dtype: binary
    - name: json
      struct:
        - name: categories
          dtype: string
        - name: description
          dtype: string
        - name: file_name
          dtype: string
        - name: label
          sequence: int32

My Emotion Video Dataset

This dataset contains short video clips annotated with emotion categories. It is saved in .parquet format, with each row containing:

a video (mp4) in binary
a json object including:
- file_name: the name of the video
- label: list of integer emotion labels
- categories: emotion name (e.g., "happy", "neutral")
- description: short description of the clip

Example usage

from datasets import load_dataset

ds = load_dataset("ZebangCheng/test_hf_data_6", split="train")
print(ds[0]['json'])
# {'categories': 'neutral', 'description': "It's me.", 'file_name': '00019.mp4', 'label': [0]}