Saved in:
Bibliographic Details
Main Authors: Dinh, Duc-Tai, Dinh, Duc Anh Khoa
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.20564
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916867134717952
author Dinh, Duc-Tai
Dinh, Duc Anh Khoa
author_facet Dinh, Duc-Tai
Dinh, Duc Anh Khoa
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2507_20564
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ZSE-Cap: A Zero-Shot Ensemble for Image Retrieval and Prompt-Guided Captioning
Dinh, Duc-Tai
Dinh, Duc Anh Khoa
Computation and Language
Information Retrieval
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.
title ZSE-Cap: A Zero-Shot Ensemble for Image Retrieval and Prompt-Guided Captioning
topic Computation and Language
Information Retrieval
url https://arxiv.org/abs/2507.20564