Doubt about non-max suppression in YOLO

In the videos, Andrew mentioned that non-max suppression must be applied independently to each class. However in Programming Assignment -1, Exercise 3, tensorflow’s non-max suppression function does not take classes as input
tf.image.non_max_suppression(
boxes,
scores,
max_output_size,
iou_threshold=0.5,
score_threshold=float(‘-inf’),
name=None
)

This has been discussed previously on the forum, and we were not able to resolve why.
So consider it an open issue the course staff is investigating.

Hello!

To my knowledge, we have separate boxes and scores for every class, right? So, if we are taking the box and score independently, we are automatically taking the class as well.

The documentation of the tf.image.non_max_suppression stated that:

Greedily selects a subset of bounding boxes in descending order of score.

Also, in the assignment, it is mentioned that:

Use tf.image.non_max_suppression() to get the list of indices corresponding to boxes you keep.

We keep separate bounding boxes for each class. Am I right?

Best,
Saif.