Closed masoud-khalilian closed 10 months ago
In order to minimize the effect of class imbalance in our model. The following link has a solution to class imbalance ENet - Real-Time Semantic Segmentation
nn.CrossEntropyLoss(**weight=torch.FloatTensor(class_weights).to(device)**) nn.CrossEntropyLoss().cuda()
In order to minimize the effect of class imbalance in our model. The following link has a solution to class imbalance ENet - Real-Time Semantic Segmentation