A Promptless Circle Search: An Innovative Approach for Efficient Query-Free Retrieval

In today's fast-paced digital world, intuitive search solutions are essential for enhancing user interaction. This paper presents Promptless Circle Search, a novel Android-based system that enables users to retrieve contextual information by simply circling an object on their screen—eliminating the need for manual text or voice queries. The system integrates YOLOv8 for real-time object detection and a locally stored GPT-2 model for generating descriptive, context-rich information. Unlike conventional methods, our system functions offline, ensuring privacy and accessibility. The core objective is to deliver a seamless, query-free retrieval mechanism that is efficient, accurate, and user-friendly. Experiments on a custom fruit dataset show the model achieves a high detection accuracy (mAP > 90%) and generates relevant contextual text. The system offers a lightweight yet powerful framework for visual search, with strong potential for global real-world application.