Fig. 3: An overfitting network learns no features of nanoparticles, but recognizes background noise. | npj Computational Materials

Fig. 3: An overfitting network learns no features of nanoparticles, but recognizes background noise.

From: Understanding important features of deep learning models for segmentation of high-resolution transmission electron microscopy images

Fig. 3

a The CNN output for a given image. b, c The raw activation values for layers detecting background and particles, respectively. The softmax function combines these activation maps to produce (a). The scale bar in a represents 50 nm and applies for all three images.

Back to article page