Figure 2
From: Neural networks for increased accuracy of allergenic pollen monitoring
Feature maps. (a) simplified view of the VGG16 model showing three convolutional layers. (b–d) Feature maps of Urticaceae pollen grains from the standard deviation projection created using ImageJ, that were confidently distinguished by the CNNs. (b) Urtica urens, (c) Parietaria judaica and (d) Urtica membranacea. Activation levels are indicated with white indicating high activation and black very low/no activation.
