The FOIL dataset annotation follows MS-COCO Image Caption Annotation Format : http://mscoco.org/dataset/#download { "info" : info, "images" : [image], "licenses" : [license], "annotations" : [annotation], } info { "year" : int, "version" : str, "description" : str, "contributor" : str, "url" : str, "date_created" : datetime, "contact" : str } image{ "id" : int, "width" : int, "height" : int, "file_name" : str, "license" : int, "flickr_url" : str, "coco_url" : str, "date_captured" : datetime, } license{ "id" : int, "name" : str, "url" : str, } annotation{ "id" : int, "foil_id" : int, "image_id" : int, "caption" : str, "target_word" : str, "foil_word" : str, "foil": boolean # True in case of foil caption else False # NOTE : target_word/foil_word = 'ORIG' in case of original/correct caption in testset. target_word/foil_word = '' (Empty) in training set. }