Struggling to understand how mAP is calculated

vision
1 
from torch import tensor
from torchmetrics.detection import MeanAveragePrecision
preds = [
  dict(
    boxes=tensor([[215.0, 41.0, 562.0, 285.0], [214.0, 41.0, 562.0, 285.0]]),
    scores=tensor([0.99, 0.09]),
    labels=tensor([0, 1]),
  )
]
target = [
  dict(
    boxes=tensor([[214.0, 41.0, 562.0, 285.0]]),
    labels=tensor([0]),
  )
]
metric = MeanAveragePrecision(iou_type="bbox", class_metrics=True, iou_thresholds=[0.5, 0.75], average="micro")
metric.update(preds, target)
from pprint import pprint
pprint(metric.compute())

This code gives:

{'classes': tensor([0, 1], dtype=torch.int32),
 'map': tensor(1.),
 'map_50': tensor(1.),
 'map_75': tensor(1.),
 'map_large': tensor(1.),
 'map_medium': tensor(-1.),
 'map_per_class': tensor([ 1., -1.]),
 'map_small': tensor(-1.),
 'mar_1': tensor(1.),
 'mar_10': tensor(1.),
 'mar_100': tensor(1.),
 'mar_100_per_class': tensor([ 1., -1.]),
 'mar_large': tensor(1.),
 'mar_medium': tensor(-1.),
 'mar_small': tensor(-1.)}

But according to my understanding, should it not give 0.5?

I am simulating two preds masks of 2 seperate classes having 100% overlap with one of the classes.

2

Another Case which should give 0, notice no classes are matched.

from torch import tensor
from torchmetrics.detection import MeanAveragePrecision
preds = [
  dict(
    boxes=tensor([[215.0, 41.0, 562.0, 285.0], [214.0, 41.0, 562.0, 285.0]]),
    scores=tensor([0.99, 0.09]),
    labels=tensor([6, 7]),
  )
]
target = [
  dict(
    boxes=tensor([[214.0, 41.0, 562.0, 285.0]]),
    labels=tensor([0]),
  )
]
metric = MeanAveragePrecision(iou_type="bbox", class_metrics=True, iou_thresholds=[0.5, 0.75], average="micro")
metric.update(preds, target)
from pprint import pprint
pprint(metric.compute())

Gives:

{'classes': tensor([0, 6, 7], dtype=torch.int32),
 'map': tensor(1.),
 'map_50': tensor(1.),
 'map_75': tensor(1.),
 'map_large': tensor(1.),
 'map_medium': tensor(-1.),
 'map_per_class': tensor([ 0., -1., -1.]),
 'map_small': tensor(-1.),
 'mar_1': tensor(1.),
 'mar_10': tensor(1.),
 'mar_100': tensor(1.),
 'mar_100_per_class': tensor([ 0., -1., -1.]),
 'mar_large': tensor(1.),
 'mar_medium': tensor(-1.),
 'mar_small': tensor(-1.)}