The success of any forest restoration project depends on the quality of the seeds used. However, for many Amazonian tree species, we lack critical information on seed viability—the ability of a seed to germinate and grow. While healthy-looking seeds with intact seed coats and high germination rates are generally considered viable, more accurate and efficient methods for assessing seed quality are needed. This project will use advanced imaging techniques, including X-rays and scanners, to analyze the internal structure of seeds from at least 200 native Amazonian tree species of economic and ecological importance. We will identify visual patterns associated with seed viability by comparing these images with germination test results. We will train a model to predict seed viability from these images using machine learning algorithms. This innovative approach will significantly reduce the time required to assess seed quality and streamline the seed production chain.