Visual Commonsense Reasoning
Visual Commonsense Reasoning (VCR) is a large-scale dataset for cognition-level visual understanding. Given a challenging question about an image, machines need to present two sub-tasks: answer correctly and provide a rationale justifying its answer. The VCR dataset contains over 212K (training), 26K (validation) and 25K (testing) questions, answers and rationales derived from 110K movie scenes.
Source: Visual Commonsense R-CNN
Image Source: From Recognition to Cognition: Visual Commonsense Reasoning
Variants: VCR, VCR (QA-R) test, VCR (QA-R) dev, VCR (Q-AR) test, VCR (Q-AR) dev, VCR (Q-A) test, VCR (Q-A) dev
This dataset is used in 1 benchmark:
| Task | Model | Paper | Date |
|---|---|---|---|
| Explanation Generation | OFA-X-MT | Harnessing the Power of Multi-Task … | 2022-12-08 |
| Explanation Generation | OFA-X | Harnessing the Power of Multi-Task … | 2022-12-08 |
Recent papers with results on this dataset: