ACRE (Abstract Causal REasoning)

摘要 因果推理（ACRE）是一个数据集，用于系统评估当前视觉系统的因果归纳，即识别导致变量之间可观察关系的不可观察机制。数据集的每个拆分结构如下： config/ train.json val.json test.json images/ ACRE_train_00*.png ACRE_val_00*.png ACRE_test_00*.png 场景/ ACRE_train_00*.json ACRE_val_00*.json ACRE_test_00*.json 每个images文件夹中的image文件在同名的场景中有对应的场景描述文件（扩展名除外）。每个 ACRE 问题都以 ACRE_{train/val/test}_{6_digit_problem_idx}_{2_digit_panel_idx} 命名

Abstract: Causal Reasoning (ACRE) is a dataset designed to systematically evaluate the causal induction of current computer vision systems, i.e., identifying the unobservable mechanisms that give rise to the observable relationships between variables.

Each dataset split follows the structure below:
config/
train.json
val.json
test.json
images/
ACRE_train_00*.png
ACRE_val_00*.png
ACRE_test_00*.png
scenes/
ACRE_train_00*.json
ACRE_val_00*.json
ACRE_test_00*.json

Each image file in the images folder has a corresponding scene description file with the same name, excluding the file extension. Each ACRE problem follows the naming convention: ACRE_{train/val/test}_{6_digit_problem_idx}_{2_digit_panel_idx}

搜集汇总

背景与挑战

背景概述

ACRE是一个用于评估视觉系统因果归纳能力的数据集，包含训练、验证和测试三个拆分，每个拆分包含图像文件和对应的场景描述文件。数据集由加州大学洛杉矶分校于2021年发布，相关论文和发布首页也已提供。