Skip to content
ml-recall

#hard-negative-mining

1 question from Object Detection

Timed test

1 question · 1 min · score and explanations at the end

Take a test
Q 1 / 1Score 0 / 0

Question 1: Short answer

easyAI-checkedObject Detection

Most anchor boxes in a detector are background (negatives). Instead of using all of them, SSD ranks the negatives by their classification loss and keeps only the hardest ones, so that negatives outnumber positives by at most 3 to 1. An image has 8,732 anchors, 8 of them positive. How many negative anchors contribute to the classification loss?

a whole number