Abstract Image Classification Dataset

本数据集名为‘Abstract Image Classification Dataset’，由因斯布鲁克大学创建，旨在通过抽象类别的图像分类挑战深度学习模型。数据集包含两个主要任务：对称性和身份识别，涉及红色棋子在棋盘上的排列。数据集通过3D建模软件Blender自动生成，可快速产生大量训练和测试图像。该数据集的应用领域在于探索和解决深度学习模型在处理抽象视觉任务时的局限性，特别是在对称性和身份识别方面的挑战。

This dataset, named "Abstract Image Classification Dataset", was developed by the University of Innsbruck. It is designed to challenge deep learning models via image classification tasks involving abstract categories. The dataset encompasses two core tasks: symmetry and identity recognition, which revolve around the arrangement of red chess pieces on a chessboard. Automatically generated using the 3D modeling software Blender, it can rapidly produce large-scale training and test image datasets. The primary application scope of this dataset is to explore and mitigate the limitations of deep learning models when processing abstract visual tasks, particularly the challenges associated with symmetry and identity recognition.