Skip to main content
. 2021 Jan 12;16(1):e0241528. doi: 10.1371/journal.pone.0241528

Fig 2. Architecture of the PlantU-net network.

Fig 2

The input is a 256×256×3 image. The hidden layer of the network includes downsampling (left) and upsampling stages (right). Both stages comprise convolution (Conv), activation (Leaky ReLU), and max pooling operations. The output is a 256 × 256 × 1 segmented image.