<![CDATA[Evaluation plays a significant role in the learning process, including the use of exams as a means of assessing students' understanding of the taught material. Continous evaluation processes assist teachers in understanding the strengths and limitations of their teaching methods. However, the importance of addressing academic dishonesty issues in evaluations is also emphasized. Upholding evaluation integrity involves security measures such as implementing online proctoring and utilizing artificial intelligence technologies. This research aims to develop an object detection-based online exam system, with the goal of enhancing evaluation effectiveness and mitigating potential cheating. The study employs the Research and Development (R&D) methodology, using the Rapid Application Development (RAD) model for application development. Various stages are undertaken in this research, starting with data collection through literature review, interviews, and observations to gather relevant information. Subsequently, the identified issues are formulated based on the collected data, followed by application modeling and development, including system requirement analysis, application process modeling, database design, and user interface development. Application testing is performed using black-box testing to ensure the quality of the final outcome, and object detection is validated using accuracy and intersection over union metrics. Expert validation is conducted to evaluate the feasibility of the generated exam system. The application is then implemented through exams administered to students, followed by data analysis based on user feedback and teacher assessment. Research findings indicate that the object detection-based online exam system positively contributes to reducing academic dishonesty by 73.1%. Assessment using the PSSUQ instrument also shows positive results for the exam system, with scores for each aspect below the maximum threshold according to PSSUQ assessment criteria.]]>
