Why the class index is different from modelnet40 ?

[cuge1995]

all_classes = ['airplane', 'bathtub', 'bed', 'bench', 'bookshelf', 'bottle',
               'bowl', 'car', 'chair', 'cone', 'cup', 'curtain', 'desk', 'door',
               'dresser', 'flower_pot', 'glass_box', 'guitar', 'keyboard',
               'lamp', 'laptop', 'mantel', 'monitor', 'night_stand', 'person',
               'piano', 'plant', 'radio', 'range_hood', 'sink', 'sofa',
               'stairs', 'stool', 'table', 'tent', 'toilet', 'tv_stand', 'vase',
               'wardrobe', 'xbox']

but in your code:

type_to_index_map = {
    'night_stand': 0, 'range_hood': 1, 'plant': 2, 'chair': 3, 'tent': 4,
    'curtain': 5, 'piano': 6, 'dresser': 7, 'desk': 8, 'bed': 9,
    'sink': 10,  'laptop':11, 'flower_pot': 12, 'car': 13, 'stool': 14,
    'vase': 15, 'monitor': 16, 'airplane': 17, 'stairs': 18, 'glass_box': 19,
    'bottle': 20, 'guitar': 21, 'cone': 22,  'toilet': 23, 'bathtub': 24,
    'wardrobe': 25, 'radio': 26,  'person': 27, 'xbox': 28, 'bowl': 29,
    'cup': 30, 'door': 31,  'tv_stand': 32,  'mantel': 33, 'sofa': 34,
    'keyboard': 35, 'bookshelf': 36,  'bench': 37, 'table': 38, 'lamp': 39
}

[thss15fyt]

The original class index is in dictionary order, while our index depends on the order of function os.listdir(). If you wanted to load our pretrained model, the referred type_to_index_map should be used. Otherwise, you could use any order of class index.