VCR

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

ImagesTexts English

벤치마크

Visual Question Answering (VQA) on VCR (Q-A) test 결과 11개

Visual Question Answering (VQA) on VCR (QA-R) test 결과 8개

Visual Question Answering (VQA) on VCR (Q-AR) test 결과 7개

Visual Question Answering (VQA) on VCR (Q-A) dev 결과 3개

Visual Question Answering (VQA) on VCR (Q-AR) dev 결과 3개

Visual Question Answering (VQA) on VCR (QA-R) dev 결과 3개

Explanation Generation on VCR 결과 2개

Visual Commonsense Reasoning on VCR (Q-A) dev 결과 1개

Visual Commonsense Reasoning on VCR (Q-A) test 결과 1개

Visual Commonsense Reasoning on VCR (Q-AR) dev 결과 1개

Visual Commonsense Reasoning on VCR (Q-AR) test 결과 1개

Visual Commonsense Reasoning on VCR (QA-R) dev 결과 1개

Visual Commonsense Reasoning on VCR (QA-R) test 결과 1개