From f7a5a57245c8978be65cf52e099075c54fd762a2 Mon Sep 17 00:00:00 2001 From: "Rezanezhad, Vahid" Date: Thu, 5 Dec 2019 16:11:31 +0100 Subject: [PATCH] Update README --- README | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/README b/README index 54ea408..e103b0b 100644 --- a/README +++ b/README @@ -1,8 +1,8 @@ -how to train: +# Train just run: python train.py with config_params.json -format of ground truth: +# Ground truth format Lables for each pixel is identified by a number . So if you have a binary case n_classes should be set to 2 and labels should be 0 and 1 for each class and pixel. @@ -15,11 +15,11 @@ format of ground truth: Label: [ [[1 0 0 1], [1 0 0 1] ,[1 0 0 1]], [[1 0 0 1], [1 0 0 1] ,[1 0 0 1]] ,[[1 0 0 1], [1 0 0 1] ,[1 0 0 1]] ] this means that you have an image by 3*4*3 and pixel[0,0] belongs to class 1 and pixel[0,1] to class 0. -training , evaluation and output: +# Training , evaluation and output train and evaluation folder should have subfolder of images and labels. And output folder should be empty folder which the output model will be written there. -patches: +# Patches if you want to train your model with patches, the height and width of patches should be defined and also number of batchs (how many patches should be seen by model by each iteration).