MUSIA-2025

Multilingual Story Illustration: Bridging Cultures through AI Artistry

Results

Results

Evaluation Criteria

  • Consistency is given the highest weight, followed by Relevance, and then Visual Quality.
  • For each criterion, the ratings follow the order: Good > Moderate > Fair.
  • The numbers in the tables represent the count of stories where the generated images fall into these categories.

Example: In Retriever-run1 (English), under Consistency, we see Good = 27, Moderate = 3, and Fair = 0. This means out of 30 stories, images for 27 stories were highly consistent, 3 were moderately consistent, and none were judged as fair.

English Results

English Results

Hindi Results

Hindi Results

Contact us

For any queries write to us at irel.iitbhu@gmail.com