Here's how the coordinates are getting assigned in build_bbox_tensors of utils/dataset.py:
bbox = info["bounding_box"]
x = bbox.get("top_left_x", bbox["topLeftX"])
y = bbox.get("top_left_y", bbox["topLeftY"])
width = bbox["width"]
height = bbox["height"]
coord_tensor[idx][0] = x
coord_tensor[idx][1] = y
coord_tensor[idx][2] = x + width
coord_tensor[idx][3] = y + height
Shouldn't this becoord_tensor[idx][3] = y - heightrather than coord_tensor[idx][3] = y + height. Assuming the xyxy format is for top left and bottom right. Also, y + height will give y coordinates that are out of the bbox. Am I missing something here?
Here's how the coordinates are getting assigned in build_bbox_tensors of utils/dataset.py:
Shouldn't this be
coord_tensor[idx][3] = y - height
rather thancoord_tensor[idx][3] = y + height
. Assuming the xyxy format is for top left and bottom right. Also, y + height will give y coordinates that are out of the bbox. Am I missing something here?