List of Tables 3.1 Number of data points per defect type for the 3D-printed products dataset. ................................ 35 3.2 Lung nodule dataset after data augmentation. . . . . . . . . . . . 38 3.3 auROCscoresforMNIST. ...................... 41 3.4 auPRCscoresforMNIST........................ 41 3.5 auROC scores for 3D-Printed Products. . . . . . . . . . . . . . . . 44 3.6 auPRC scores for 3D-Printed Products. . . . . . . . . . . . . . . . 44 4.1 Encoder and decoder architectures used in most methods. . . . . 78 4.2 Encoder and decoder architecture used to train LSBD-VAE/0 for ModelNet40dataset.......................... 78 4.3 LSBD-VAE hyperparameters for all datasets. . . . . . . . . . . . . 79 4.4 Model hyperparameters for all datasets. . . . . . . . . . . . . . . 79 4.5 ScoresfortheSquaredataset. . . . . . . . . . . . . . . . . . . . . 85 4.6 ScoresfortheArrowdataset. . . . . . . . . . . . . . . . . . . . . 86 4.7 Scores for the Airplane dataset. . . . . . . . . . . . . . . . . . . . 87 4.8 Scores for the ModelNet40 dataset. . . . . . . . . . . . . . . . . . 88 4.9 ScoresforCOIL-100dataset.. . . . . . . . . . . . . . . . . . . . . 88 4.10 Hyperparameters for the different variants submitted to the SHREC 2021 3D object retrieval challenge. . . . . . . . . . . . . . . . . . 100 4.11 Evaluation measures for the retrieval-by-shape challenge. . . . . 101 4.12 Evaluation measures for the retrieval-by-culture challenge. . . . . 102 5.1 OOD splits for dSprites and 3D Shapes. . . . . . . . . . . . . . . . 112
RkJQdWJsaXNoZXIy MjY0ODMw