Week 7 - Deep Learning Architectures & Training

Question 1

What is the concept of Transfer Learning?

Answer 1

Transfer Learning is a method that leverages a pre-trained network for new tasks without needing vast amounts of new training data.

Question 2

How does Transfer Learning work?

Answer 2

• A network is initially trained on a large, general-purpose dataset like ImageNet
• Then, if the specific dataset is too small, you freeze the pre-trained network’s weights and only re-train the classifier layer
• For a medium sized dataset, you start with the pre-trained weights, then re-train the whole network or just the higher layers using a reduced learning rate.

Question 3

What are some properties of the ImageNet dataset?

Answer 3

• 1.2 million high-resolution images
• 1000 different classes

Question 4

What are some historical examples of Classification CNNs?

Answer 4

• AlexNet
• VGG
• Residual Network

Question 5

What are some historical examples of Segmentation CNNs?

Answer 5

• U-Net
• SegNet

Question 6

What are some historical examples of Object Detection CNNs?

Answer 6

• RCNN
• Fast RCNN
• Mask RCNN
• YOLO
• nnU-Net

Question 7

How does Eye Tracking get used effectively in Computer Vision?

Answer 7

Deep learning performs better when regions are marked before the learning process.
As such, eye tracking is a cheap and easy way to record where someone has been looking. Once this data is filtered, it can be passed to a suitable deep learning system.

Question 8

What can help to ‘pad-out’ a dataset if it isn’t suitably large enough for training?

Answer 8

Creating and using synthetic data, as it can be used to augment the original dataset with ‘new’ data

Question 9

What are some factors to consider when rendering synthetic images?

Answer 9

• Accuracy of material
• Lighting
• Background
• How to create annotations