def _sliding_window_processor(engine, batch):
net.eval()
img, seg, meta_data = batch
with torch.no_grad():
seg_probs = sliding_window_inference(img, roi_size, sw_batch_size, net,
device)
return seg_probs, seg.to(device)
def _sliding_window_processor(_engine, batch):
net.eval()
img, seg, meta_data = batch
with torch.no_grad():
seg_probs = sliding_window_inference(img, roi_size, sw_batch_size,
lambda x: net(x)[0], device)
return predict_segmentation(seg_probs)
def test_sliding_window_default(self, image_shape, roi_shape, sw_batch_size):
inputs = np.ones(image_shape)
device = torch.device("cpu:0")
def compute(data):
return data.to(device) + 1
result = sliding_window_inference(inputs, roi_shape, sw_batch_size, compute, device)
expected_val = np.ones(image_shape, dtype=np.float32) + 1
self.assertTrue(np.allclose(result.numpy(), expected_val))
def _sliding_window_processor(engine, batch):
net.eval()
with torch.no_grad():
seg_probs = sliding_window_inference(batch['img'], roi_size,
sw_batch_size, net, device)
return seg_probs, batch['seg'].to(device)