dnamodel/adcumen-viewer-emotions

AdCumen Viewer Emotions Dataset

Dataset for the paper "Decoding Viewer Emotions in Video Ads" by Alexey Antonov, Shravan Sampath Kumar, Jiefei Wei, William Headley, Orlando Wood, and Giovanni Montana, published in Nature Scientific Reports.

Code: github.com/gmontana/DecodingViewerEmotions Model weights: dnamodel/tsam-viewer-emotions

Dataset Description

The dataset consists of 26,637 five-second video clips extracted from video advertisements, annotated for seven distinct emotional categories and their temporal onset. The clips were derived from System1's "Test Your Ad" tool, with each original advertisement annotated by approximately 75 viewers.

Emotion Labels

ID	Emotion
0	Anger
1	Contempt
2	Disgust
3	Fear
4	Happiness
5	Sadness
6	Surprise

Dataset Splits

Emotion	Total	Train	Validation	Test
Anger	2,894	2,282	404	208
Contempt	3,317	2,581	367	369
Disgust	3,061	2,564	254	243
Fear	3,166	2,549	317	300
Happiness	3,577	2,918	383	276
Sadness	3,576	2,886	346	344
Surprise	3,553	2,841	387	325
Total	26,637	21,392	2,856	2,387

Files

File	Description
training.csv	Training split (21,392 clips)
validation.csv	Validation split (2,856 clips)
testing.csv	Test split (2,387 clips)
5-second_MP4_Clips.zip	All 5-second video clips (MP4)

Each CSV contains:

Column	Description
Video_Name	Unique video clip identifier
Start_Second	Start time in the original advertisement
Label	Emotion label (0–6)
Clips_Name	Clip filename

Usage

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="dnamodel/adcumen-viewer-emotions",
    repo_type="dataset",
    local_dir="./adcumen-data"
)

See the code repository for full training and inference instructions.

Citation

@article{antonov2024decoding,
  title={Decoding viewer emotions in video ads},
  author={Antonov, Alexey and Kumar, Shravan Sampath and Wei, Jiefei and Headley, William and Wood, Orlando and Montana, Giovanni},
  journal={Scientific Reports},
  volume={14},
  pages={25680},
  year={2024},
  publisher={Nature Publishing Group},
  doi={10.1038/s41598-024-76968-9}
}

License

The dataset is provided under a custom license from the University of Warwick. Use is permitted solely for academic research and non-commercial evaluation. See the LICENSE file for full terms.

The dataset leverages System1's proprietary "Test Your Ad" tool for public, educational, and illustrative use. The advertisements and excerpts remain the property of their original owners. Usage beyond this study's scope requires explicit permission from those owners.

Contact

• Questions or collaborations: Giovanni Montana — g.montana@warwick.ac.uk
• Commercial licensing: Warwick Ventures — ventures@warwick.ac.uk