Semantic Segmentation with Amazon Sagemaker
Graded Quiz: Test your Project Understanding
1.
Question 1
We used a high level Sagemaker Estimator to create, train and deploy our Semantic Segmentation model instead of implementing the algorithm ourselves. True or False?
2.
Question 2
Which channels do we need to specify in data channels when using the Sagemaker Semantic Segmentation with ‘File’ input mode that we used in the hands on project? Select all that apply:
3.
Question 3
We used GPUs on our Notebook Instance, the one we used to prepare data and Estimator to launch the training job, as well as on the instance launched to train the model. True or False?
Correct! We did not use a GPU on the Notebook instance since we did not need to train any model on it. The training is done on a separate instance launched by the Estimator which used a GPU.
================================================
4.
Question 4
Deployed model endpoint is deleted automatically when the Notebook instance is stopped. True or False?
Correct! You have to explicitly delete the endpoint.
================================================
5.
Question 5
The deployed model categorizes input image pixels in one of three classes: Foreground, Background, Not classified or Border. True or False?
Correct, the prediction is a segmentation map like our annotations. All pixels of the input image are assigned a value of either 1, 2 or 3.