You are working as a machine learning engineer for an online retail company that is developing a visual search engine to identify whether an image contains one of the company's products. You have implemented an end-to-end ML pipeline on Google Cloud to handle this task. With new product releases anticipated, you have set up a retraining functionality in the pipeline to incorporate new data into your ML models. Additionally, you aim to utilize AI Platform's continuous evaluation service to maintain high accuracy levels on the test dataset. Considering these requirements, how should you manage your test dataset when integrating images of new products into the retraining process?