Closed Strike1999 closed 2 years ago
Thanks for bringing this up! PhraseCut.tar
is not directly available but must be composed by downloading the PhraseCut dataset. PhraseCut.tar should have this structure:
VGPhraseCut_v0/name_att_rel_count.json
VGPhraseCut_v0/image_data_split.json
VGPhraseCut_v0/refer_miniv.json
VGPhraseCut_v0/refer_input_val.json
VGPhraseCut_v0/refer_test.json
VGPhraseCut_v0/refer_train.json
VGPhraseCut_v0/refer_input_miniv.json
VGPhraseCut_v0/refer_input_test.json
VGPhraseCut_v0/refer_val.json
VGPhraseCut_v0/refer_input_train.json
VGPhraseCut_v0/images/... <- here are the images from Visual Genome
Thanks for your great job!
How can I get PhraseCut.tar in
phrasecut.py
?