Food/Non-food Image Classification

Recently food recognition is a vital topic among researchers, as it automates the food logging process besides having applications in Food Safety, Food Quality Detection, Inspection, etc. The advancement in deep learning technology on edge computing and cloud computing leads to the development of hybrid solutions by considering the advantages of these computing platforms. For large scale, experiments, we gathered and prepared a Food/Non-Food dataset created from the existing image datasets of UECFOOD256, Caltech 256, Instagram Images, Flickr Image Dataset, Food101, Malaysian Food Dataset(gathered and crawled by us), Indoor Scene recognition Dataset, 15 scene dataset. The dataset is divided into two classes and is currently the largest Food/NonFood dataset available to the best of our knowledge. Our main contribution for this repository is the Food/Non-Food dataset, using foods of different cultures, existing food datasets, and the images of the users on the popular social media platforms. We manually cleaned the datasets for this purpose.