PICS: Pipeline for Image Captioning and Search

Grant Rosario,David Noever

2024-02-01

Abstract:The growing volume of digital images necessitates advanced systems for efficient categorization and retrieval, presenting a significant challenge in database management and information retrieval. This paper introduces PICS (Pipeline for Image Captioning and Search), a novel approach designed to address the complexities inherent in organizing large-scale image repositories. PICS leverages the advancements in Large Language Models (LLMs) to automate the process of image captioning, offering a solution that transcends traditional manual annotation methods. The approach is rooted in the understanding that meaningful, AI-generated captions can significantly enhance the searchability and accessibility of images in large databases. By integrating sentiment analysis into the pipeline, PICS further enriches the metadata, enabling nuanced searches that extend beyond basic descriptors. This methodology not only simplifies the task of managing vast image collections but also sets a new precedent for accuracy and efficiency in image retrieval. The significance of PICS lies in its potential to transform image database systems, harnessing the power of machine learning and natural language processing to meet the demands of modern digital asset management.

Computer Vision and Pattern Recognition,Information Retrieval,Machine Learning