Semantic Segmentation and the Dataset

https://d2l.ai/chapter_computer-vision/semantic-segmentation-and-dataset.html

Thanks for the VOC2012 segmentation dataloader template. I didn’t expect to find a detail dataloader explanation from this book at the beginning.

Thanks for the explanation. However, when I load the dataset, there are no labels with category number 8-15 and 20 i.e

[ ‘cat’, ‘chair’, ‘cow’, ‘diningtable’, ‘dog’, ‘horse’, ‘motorbike’, ‘person’] and [‘train’]

This is what I got when I counted the number of instances for each category in the train dataset:

[182871358, 2031356, 713150, 4989269, 2301218, 1463550, 2336871, 3844779, 0, 0, 0, 0, 0, 0, 0, 0, 1428974, 2114231, 3422716, 0, 2561556]

Similarly, indices 8-15 and 20 are also 0 in the validation dataset. Could anyone explain to me why is it so?