Developed and delivered an image object coordinate localization feature for the XiaoMi/mone repository, enabling automated identification of object positions within images using a large language model. The implementation centered on Java, leveraging both image processing and LLM integration to infer coordinates from sample images. The workflow included the creation of LocateCoordinates.java and a dedicated unit test, LocateCoordinateTest.java, which validated the end-to-end localization process. This work established a foundation for downstream tasks such as tagging, cropping, and analytics, while emphasizing a test-driven approach and early integration readiness. No major bug fixes were addressed, as the focus remained on robust feature delivery.