Visual Question Answering v2.0 (VQA v2.0)
Introduced by Goyal et al. in Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering
Visual Question Answering (VQA) v2.0 is a dataset containing open-ended questions about images. These questions require an understanding of vision, language and commonsense knowledge to answer. It is the second version of the VQA dataset.
- 265,016 images (COCO and abstract scenes)
- At least 3 questions (5.4 questions on average) per image
- 10 ground truth answers per question
- 3 plausible (but likely incorrect) answers per question
- Automatic evaluation metric
The first version of the dataset was released in October 2015.
Benchmarks
| Trend | Task | Dataset Variant | Best Model | Paper | Code |
|---|---|---|---|---|---|
|
VQA v2 test-dev
|
PaLI
|
|||
|
VQA v2 test-std
|
BEiT-3
|
|||
|
VQA v2 val
|
BLIP-2 ViT-G FlanT5 XXL
|
|||
|
VQA v2 test-dev
|
mPLUG
|
|||
|
VQA v2 val
|
BLIP-2 ViT-G OPT 6.7B
|
|||
|
VQA v2 test-std
|
Aurora
|
|||
|
VQA v2
|
LXMERT
|
Papers
| Paper | Code | Results | Date | Stars |
|---|
