ImageEval SharedTask: Baselines and Tutorial for Arabic Image Captioning (Subtask 2)

[Mustafa Jarrar]

يسعدنا أن نشارككم البرامج والبيانات التدريبية لنموذج وصف الصور– تحتاجون بضع ساعات للوصول الى الدقة المرجعية. 

ندعوكم للمشاركة في المسابقة البحثية.

Dear All,

We are happy to share this tutorial for the ImageEval Shared Task (Colab notebooks + dataset) on Arabic image captioning (Subtask 2). These results are the baseline.

These baselines are now available on both GitHub and Hugging Face to support participants to get started with training and benchmarking their models easily - whether for zero-shot evaluation or fine-tuning. We encourage you to explore these resources and participate in the shared task.

🔗 Link: https://sina.birzeit.edu/image_eval2025/

 

Note: We also encourage you to participate in Subtask 1 (the datathon), which involves manually captioning images.

 

Best Regards,
—Mustafa Jarrar