Hello, I’m stuck on “Exercise 1 - yolo_filter_boxes”, specifically subparts 3 and 4. Could it be possible for some to explain to me what exactly I’m supposed to be doing? How do I filter mask (I’ve read the documentation on the TF page, I still don’t quite follow why this needs to be done)?
Could someone help me please?
Thank you!
[code removed - moderator]
Here are some hints:
Step 2:
The last dimension holds the scores per class. box_classes is created by picking the class that has the highest score.
box_class_scores should be these maximum scores based on which box_classes was initialized. If you don’t specify an axis, tf.math.reduce_max picks a single maximum value across all dimensions. This is not what you want. Similar to box_classes, pick the max score among the 80 classes.
Step 3:
filtering_mask should be the same shape as box_class_scores. Each entry in this mask should be True if the entry in box_class_scores > threshold, False otherwise.