Question about the newest implementation of tf.image.non_max_suppression

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

In the Autonomous_driving_application_Car_detection programming assignment (week 3), in exercise 3, we use an older version of tf.image.non_max_suppression() which doesn’t take in score_threshold as parameter; however, the latest version does.

Does this new version of non max supression apply the filtering of boxes by class score? I.e., does this new version of tf.image.non_max_suppression() include in its implementation what we implemented as yolo_filter_boxes()? (Exercise 1 of the Autonomous_driving_application_Car_detection programming assignment)

2

As far as I can see it does part of it; it computes the filtering_mask but the tf.image.non_max_suppression() still needs for example the boxes parameter as comes out of the yolo_filter_boxes() function not the one entering this function as parameter. SO I dont think you can entirely remove that function as it is right now.