rmi-pytorch 0.1.1

Region Mutual Information loss in PyTorch

PyTorch implementation of the Region Mutual Information Loss for Semantic Segmentation.

Example usage

With logits:

import torch
from rmi import RMILoss

loss = RMILoss(with_logits=True)

batch_size, classes, height, width = 5, 4, 64, 64
pred = torch.rand(batch_size, classes, height, width, requires_grad=True)
target = torch.empty(batch_size, classes, height, width).random_(2)

output = loss(pred, target)
output.backward()

With probabilities:

import torch
from torch import nn
from rmi import RMILoss

m = nn.Sigmoid()
loss = RMILoss(with_logits=False)

batch_size, classes, height, width = 5, 4, 64, 64
pred = torch.randn(batch_size, classes, height, width, requires_grad=True)
target = torch.empty(batch_size, classes, height, width).random_(2)

output = loss(m(pred), target)
output.backward()

Graphs

Plot of the value of the loss between the prediction and target without the BCE component. Target is a random binary 256x256 matrix. For Random the prediction is a 256x256 matrix of probabilities initialized uniformly at random. For All zero the prediction is a 256x256 matrix with all zeros. For 1- target the prediction is the inverse of the target. The prediction is interpolated with the target by: input_i = (1 - α) * input + α * target.

Difference between this implementation and the implementation in the official git repository, with EPSILON = 0.0005 and pool='max'.

Execution time on tensors with batch size of 8 and with 21 classes.

Size	This	Official
8x21x32x32	6.5722ms	6.3261ms
8x21x64x64	11.8159ms	12.6169ms
8x21x128x128	39.9946ms	40.3798ms
8x21x256x256	160.0352ms	160.9543ms