Problem visualizing the 2D bounding boxes

Hello,
I have used your script to convert 3D bboxes into 2D ones (python-sdk/scripts/export_2d_annotations_as_json). I have setted the visibilities equals to 2,3 and 4 because i would like that the objects occluded under 40% are excluded. Besides i have retrieved the images from each camera. After that, i have cropped the images and i have visualized them.
It seems that a certain number of images are visible less than the 40 % . Here a sample of images:

How is it possible?

Hi. Please refer to the schema at https://github.com/nutonomy/nuscenes-devkit/blob/master/schema.md#visibility .
“The visibility of an instance is the fraction of annotation visible in all 6 images”. So e.g. 30% visibility can be 10% in one image and 20% in another.