metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: question
      dtype: string
    - name: answer
      dtype: string
    - name: category
      dtype: string
  splits:
    - name: train
      num_bytes: 193480618
      num_examples: 1200
  download_size: 191167972
  dataset_size: 193480618
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: mit
task_categories:
  - visual-question-answering
language:
  - en
size_categories:
  - 1K<n<10K

VLM Benchmark for Facial Feature Understanding

To evaluate the model's ability to interpret facial features, including attributes such as emotion, age, gender, ethnicity, and user count, we employed the Face Task Bench, a benchmark comprising 1,200 entries.

The benchmark covers six distinct tasks related to facial feature analysis, including emotion and age prediction.
Each task consists of 200 diverse entries, ensuring a robust dataset for assessing the model’s capacity to interpret and analyze facial characteristics accurately.

This benchmark aids in evaluating the model’s effectiveness in understanding human facial attributes and enhances its ability to provide context-aware and fair assessments.

Citation

If you use this dataset in your research, please cite the original sources:


@article{rahimi2025user,
  title={User-VLM: LLM Contextualization with Multimodal Pre-trained User Models},
  author={Rahimi, Hamed and Abrini, Mouad and Khoramshahi, Mahdi and Chetouani, Mohamed},
  year={2025}
}

@article{rahimi2025user,
  title={User-VLM 360°: Personalized Vision Language Models with User-aware Tuning for Social Human Robot Interactions},
  author={Rahimi, Hamed and Bhaj, Adil, and Abrini, Mouad and Khoramshahi, Mahdi and Ghogho, Mounir and Chetouani, Mohamed},
  year={2025}
}

License

This dataset is licensed under the MIT License.

Contact

For any questions or issues regarding the dataset, please open an issue on this repository or contact the dataset maintainers.