mirror of
https://github.com/qurator-spk/sbb_pixelwise_segmentation.git
synced 2025-06-09 11:50:04 +02:00
Update README
This commit is contained in:
parent
036e2e9de5
commit
f7a5a57245
1 changed files with 4 additions and 4 deletions
8
README
8
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).
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue