How Do ConvNets Understand Image Intensity?
This addresses the understanding of ConvNet decision-making processes for researchers in computer vision, but it appears incremental as it extends existing visualization techniques to a specific aspect.
The paper investigates how Convolutional Neural Networks (ConvNets) rely on image intensity information in addition to edge/shape cues for classification, showing this through visualization methods.
Convolutional Neural Networks (ConvNets) usually rely on edge/shape information to classify images. Visualization methods developed over the last decade confirm that ConvNets rely on edge information. We investigate situations where the ConvNet needs to rely on image intensity in addition to shape. We show that the ConvNet relies on image intensity information using visualization.