Fig. 6
From: TasselNetv2: in-field counting of wheat spikes with context-augmented local regression networks
The structure of TasselNet, TasselNet added context and TasselNetv2. All of the networks adopt AlexNet-like architectures. The definition of the convolutional and pooling layers is in the format: fliter size + layer name, number of channels, padding, /stride. Fully connected layers are defined in the format: layer name, number of nodes. The different settings are highlighted in red