1.data processing
Multi_make_anno.py
2.neural network
Multi_make_anno.py
3.cross entropy loss(classes and speices)
classes loss weight rate:0.8, species loss weight rate:0.2
loss = 0.8 * loss_classes + 0.2 * loss_species
training epochs: 100
the best accuracy： 0.5750
Multi_classification.py
4.binary cross entroy(torch.nn.BCELoss()),use One-Hot Encoding
Multi_classification_bce.py
