Convolutional Recurrent Neural Networks for Observation-Centered Plant Identification

Figure 3

Effects of (a) ResNet50, (b) InceptionV3, (c) Xception, and (d) MobileNet CNN with three different RNN units on the 10-fold test accuracy. The pentagrams in (b), (c), and (d) represent the outlier of the results.

(a)

(b)

(c)

(d)