ZSE-Cap: A Zero-Shot Ensemble for Image Retrieval and Prompt-Guided Captioning
This work addresses the challenge of linking textual articles to images without task-specific fine-tuning, though it is incremental as it builds on existing foundation models.
The paper tackled the problem of article-grounded image retrieval and captioning by proposing a zero-shot ensemble system that combines CLIP, SigLIP, and DINOv2 for retrieval and uses a prompt-guided Gemma 3 model for captioning, achieving a final score of 0.42002 and securing 4th place in the EVENTA shared task.
We present ZSE-Cap (Zero-Shot Ensemble for Captioning), our 4th place system in Event-Enriched Image Analysis (EVENTA) shared task on article-grounded image retrieval and captioning. Our zero-shot approach requires no finetuning on the competition's data. For retrieval, we ensemble similarity scores from CLIP, SigLIP, and DINOv2. For captioning, we leverage a carefully engineered prompt to guide the Gemma 3 model, enabling it to link high-level events from the article to the visual content in the image. Our system achieved a final score of 0.42002, securing a top-4 position on the private test set, demonstrating the effectiveness of combining foundation models through ensembling and prompting. Our code is available at https://github.com/ductai05/ZSE-Cap.