Skip to content

wangpengnorman/FVQA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 

Repository files navigation

FVQA: Fact-based Visual Question Answering

It can be downloaded from here.

  • ./Name_Lists: the txt files contain the train and test images' id in the dataset.
  • ./new_dataset_release/Images: the image files used by this dataset, which are extracted from COCO and ImageNet datasets.
  • ./new_dataset_release/all_fact_triples_release.json: the facts (triplets <e1, rel, e2>) corresponding to the visual concepts shown in the paper (object/scene/action), which are extracted from three knowledge bases, i.e., DBpedia, ConceptNet and Webchild.
    -- r: the unique id of entity rel (relationship)
    -- e1: the unique id of entity e1
    -- e2: the unique id of entity e2
    -- KB: the source of knowledge base
    -- e1_label: the textual string describing entity e1
    -- e2_label: the textual string describing entity e2
    -- surface: the textual description of the fact
  • ./new_dataset_release/all_qs_dict_release.json: the collected questions and their associated supporting facts.
    -- fact_surface: the textual description of the supporting fact
    -- ans_source: the source of the answer (Image or KB)
    -- question: the collected question
    -- answer: the answer to the question
    -- img_file: the image filename corresponding to the question
    -- visual concept: the visual concept corresponding to the supporting fact
    -- fact: the unique id of the supporting fact
    -- question id: the unique id of the question

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published