Object detection: wrong class id and class name mappings in "categories" list in an image annotation file

While running the [example](https://github.com/awslabs/amazon-sagemaker-examples/blob/master/introduction_to_amazon_algorithms/object_detection_pascalvoc_coco/object_detection_image_json_format.ipynb) given by Amazon SageMaker for Object Detection, I'm not able to understand how the "categories" property stores the mapping between the class index and class name. For instance, for an image "000000183716.jpg" the annotation file "000000183716.json" generated after running the code in juypter notebook has the following contents:
`{"file": "000000183716.jpg", "image_size": [{"width": 353, "height": 500, "depth": 3}], "annotations": [
{"class_id": 27, "top": 259, "left": 86, "width": 32, "height": 41},
{"class_id": 27, "top": 229, "left": 220, "width": 51, "height": 248},
{"class_id": 0, "top": 47, "left": 2, "width": 219, "height": 447},
{"class_id": 56, "top": 74, "left": 8, "width": 340, "height": 156},
{"class_id": 0, "top": 14, "left": 148, "width": 203, "height": 468}],
"categories": [{"class_id": 32, "name": "tie"},
{"class_id": 32, "name": "tie"},
{"class_id": 1, "name": "person"},
{"class_id": 62, "name": "chair"},
{"class_id": 1, "name": "person"}]}`
In the above JSON annotations list, [27, 0, 56] corresponds to ["tie", "person", "face"] respectively. However, the categories list [32, 1, 62] corresponds to [tie, person, chair] respectively. Though, there is no chair in this image. I want to know how exactly these two properties work.