In SAS Viya's deep learning pipeline, the DLPy ImageTable.loadimage…

Question

In SAS Viya's deep learning pipeline, the DLPy ImageTable.loadimages() method loads images from a local directory into a CAS in-memory table. What additional CAS action is typically needed to convert raw loaded images to model input format?

Accepted Answer

The ImageProcessing.resize() action set (or ImageTable.resize()) must be called to resize images to the expected height/width dimensions, and optionally normalize pixel values — Here's the deal — b is correct because loadimages() loads raw image bytes into a CAS binary table (typically into a 'image' binary column), but the images may be different sizes and formats. Before training, images typically need to be resized to the model's expected input dimensions using the ImageProcessing CAS action set or DLPy's ImageTable helper methods.

Answer

No additional action is needed; loadimages() automatically resizes and normalizes images to the model's expected input dimensions

Answer

PROC CONTENTS must be run on the CAS image table to verify that image bytes loaded correctly before model training

Answer

The CAS session must be terminated and restarted after loading images to initialize GPU memory allocation

In SAS Viya's deep learning pipeline, the DLPy ImageTable.load_images() method loads images from a local directory into a CAS in-memory table. What additional CAS action is typically needed to convert raw loaded images to model input format?

Short Explanation and Infographic

Full Explanation