Аналіз моделі openai clip для задачі пошуку зображень на персональному комп’ютері за текстовим запитом

Abstract

This study examines the use of the OpenAI CLIP model for image search on personal computers, addressing the inefficiencies of traditional metadata-based methods. By utilizing a custom dataset of 11 thematic, the research explores CLIP’s ability to map images and text into a shared feature space using embeddings. The model’s effectiveness is demonstrated through its capacity to discern semantic relationships and differentiate between similar and distinct classes. The results indicate that integrating CLIP can significantly enhance media content management and search efficiency. Future work involves developing an application that combines metadata and semantic search capabilities to improve the quality and convenience of image search on personal computers.