CS 7643 / CS7643 Quiz 3 (Latest
Update ) Deep Learning
| Questions & Answers | Grade A |
100% Correct - Georgia Tech
Question:
What does R-CNN stand for?
Answer:
Region-based Convolutional Neural Network
Question:
What is the purpose of R-CNN?
Answer:
To find regions of interest (ROIs) with object-like things.
,Question:
What method does R-CNN use to find ROIs?
Answer:
Selective Search
Question:
What does R-CNN return for each ROI?
Answer:
Scores and bounding boxes
Question:
Why is R-CNN considered slow?
Answer:
It processes hundreds of images/crops for the same image portions.
Question:
What does R-CNN do after finding ROIs?
Answer:
Classifies those regions and refines their bounding boxes.
,Question:
What does Fast R-CNN map each ROI in an image to?
Answer:
Corresponding region in feature maps.
Question:
How does Fast R-CNN reuse computation?
Answer:
By finding regions in feature maps.
Question:
How many times is feature extraction performed per image in Fast R-CNN?
Answer:
Once per image.
Question:
What issue does Fast R-CNN face due to variable input size?
Answer:
Variable input size to fully connected (FC) layers.
, Question:
What technique does Fast R-CNN use to solve the variable input size issue?
Answer:
ROI Pooling.
Question:
What is Faster R-CNN used for?
Answer:
Deep Learning to do object detection.
Question:
What does RPN stand for in Faster R-CNN?
Answer:
Region Proposal Network.
Question:
What does the RPN in Faster R-CNN generate?
Answer:
Proposals for object detection.
Update ) Deep Learning
| Questions & Answers | Grade A |
100% Correct - Georgia Tech
Question:
What does R-CNN stand for?
Answer:
Region-based Convolutional Neural Network
Question:
What is the purpose of R-CNN?
Answer:
To find regions of interest (ROIs) with object-like things.
,Question:
What method does R-CNN use to find ROIs?
Answer:
Selective Search
Question:
What does R-CNN return for each ROI?
Answer:
Scores and bounding boxes
Question:
Why is R-CNN considered slow?
Answer:
It processes hundreds of images/crops for the same image portions.
Question:
What does R-CNN do after finding ROIs?
Answer:
Classifies those regions and refines their bounding boxes.
,Question:
What does Fast R-CNN map each ROI in an image to?
Answer:
Corresponding region in feature maps.
Question:
How does Fast R-CNN reuse computation?
Answer:
By finding regions in feature maps.
Question:
How many times is feature extraction performed per image in Fast R-CNN?
Answer:
Once per image.
Question:
What issue does Fast R-CNN face due to variable input size?
Answer:
Variable input size to fully connected (FC) layers.
, Question:
What technique does Fast R-CNN use to solve the variable input size issue?
Answer:
ROI Pooling.
Question:
What is Faster R-CNN used for?
Answer:
Deep Learning to do object detection.
Question:
What does RPN stand for in Faster R-CNN?
Answer:
Region Proposal Network.
Question:
What does the RPN in Faster R-CNN generate?
Answer:
Proposals for object detection.
