One of the groundbreaking achievements of the AfroPALM project was the creation of the first Afrocentric visual image dataset focused on red palm oil adulteration. This dataset was meticulously compiled by capturing approximately 5,500 high-quality images of palm oil samples under various conditions, including different storage times, container types, and adulterant exposure. The diversity and comprehensiveness of this dataset provided a rich resource for training the deep learning models, ensuring they could accurately distinguish between pure and adulterated palm oil. This dataset is a valuable asset for the AfroPALM project and a critical resource for future food safety and quality control research across Africa.
The datasets for the AfroPALM model development can be accessed here: https://drive.google.com/drive/folders/1VedA1zi5X5cgtMU1POSzISmFLvr9ROAU