fix COCO box format in dataset wrapper

[pmeier]

See this manual XYWH -> XYXY conversion in the references

vision/references/detection/coco_utils.py

Lines 60 to 63 in 9b233d4

	boxes = torch.as_tensor(boxes, dtype=torch.float32).reshape(-1, 4)
	boxes[:, 2:] += boxes[:, :2]
	boxes[:, 0::2].clamp_(min=0, max=w)
	boxes[:, 1::2].clamp_(min=0, max=h)

or this from the datasets v2:

vision/torchvision/prototype/datasets/_builtin/coco.py

Lines 128 to 132 in 9b233d4

	bounding_boxes=BoundingBox(
	[ann["bbox"] for ann in anns],
	format="xywh",
	spatial_size=spatial_size,
	),

cc @vfdev-5 @bjuncek

[NicolasHug]

Thanks Philip