This project aims to develop domain adaptation techniques for UAV (Unmanned Aerial Vehicle) imagery to enhance object detection performance under varying environmental conditions such as fog, rain, brightness changes, and low-light scenarios. It will be conducted at Yuan Ze University, Department of Computer Science, under the supervision of Qazi Mazhar Ul Haq, from 23 March 2026 to 9 August 2026 (approximately 4.5 months), exceeding the minimum two-month requirement. The program will involve four students primarily recruited from Universiti Putra Malaysia and University Muhammadiyah Surabaya, who are expected to have backgrounds in computer science, informatics, electrical engineering, or related fields, along with basic knowledge of machine learning, programming (e.g., Python), and proficiency in English for academic communication. The project is structured into three phases: research foundation (literature review and dataset analysis such as VisDrone and UAVDT), guided development (designing adaptive detection models using techniques like style transfer, adversarial learning, or self-supervised learning under close mentorship), and evaluation and reporting (experimental validation targeting 5–10% mAP improvement, technical reporting, and potential publication or prototype development). A total funding of NTD 360,000 will be provided as a student allowance, equivalent to NTD 15,000 per month per student for six months, to support living expenses; accommodation will be self-arranged by participants with optional guidance from the university, including dormitory or off-campus options. Participants are required to commit to the full duration, actively engage in research collaboration, and contribute to project outcomes, which are expected to include robust UAV detection models, strengthened international academic partnerships, and enhanced technical competencies in AI, domain adaptation, and UAV analytics.