I've fixed this bug on my branch like this: https://github.com/yli-hallila/qupath/blob/master/qupath-core/src/main/java/qupath/lib/objects/PathAnnotationObject.java#L77-L81

master bumped to 0.3.0-SNAPSHOT in 030bfd285907ee95d471cf5313001c6ddb260467.

Follow on from discussion on #78. Major version checklist:

• [ ] Bump master to 0.3.0-SNAPSHOT
• [ ] Document the removal of N5 support and merge #78
• [ ] Document changes to command line arguments
• [ ] Bump layout version
• [ ] Update layout based on suggestions from #78 (will have an effect on isyntax2raw and raw2ometiff)
• [ ] Document existing layout and changes to layout

/cc @melissalinkert, @joshmoore

Hey @maweigert,

Thank you for testing. According to the nvidia-manual:

sudo nvidia-smi -i 0 -c EXCLUSIVE_PROCESS

will set GPU 0 to exclusive process mode. I currently do not have access to a GPU-machine on which I can run as super-user, so I cannot test if

suda nvidia-smi -c 0

would do the trick, too.

Regarding 2.: No, it does not throw an error.

Best, D

Bug report

Describe the bug In v0.2 it isn't possible to set an annotation description, then reset it (by removing the text).

To Reproduce Steps to reproduce the behavior:

1. Create an annotation
2. Change the description in the properties and apply the change
3. Delete the description and (try to) apply the change

Alternatively use this script:

def roi = ROIs.createRectangleROI(0, 0, 100, 100, ImagePlane.getDefaultPlane())
def annotation = PathObjects.createAnnotationObject(roi)
annotation.setDescription("something")
annotation.setDescription(null)

Expected behavior Setting description to null or an empty string removes it.

Desktop (please complete the following information):

• QuPath Version 0.2

Additional context See https://forum.image.sc/t/unable-to-clear-annotation-description-in-qupath/47741

Should be quite straightforward to fix, but subtleties take a bit of thought.

How can I save 16 bit images sequence to amira ascii format. Works when images are 8 bits.

How about creating a new topic ImageJ FAQ on the forum, where we can group frequently asked questions like this one, and then link to it with a more concise link?

Related: we used to have https://imagej.net/FAQ as well, but it's currently still read-only, and will likely take a while before the wiki is fully migrated.

Hi @frauzufall, @ctrueden,

I would like to add a link to the forum to an error message. We are repeatedly answering that question on the forum and I think it might help to have that short-cut directly in the error message.

Let me know what you think!

Cheers, Robert

From the presentation at NEUBIAS Academy by @bogovicj:

When applying the transform to the moving image, the spatial calibration is correct, but the unit might get wrong (e.g. µm instead of nm).

Hello Stardist team,

Thank you for your work! Unfortunately, I still have a memory issue even using @maweigert proposal.

Firstly, you can see just below the memory increase that I faced:

Epoch 20 Beginning

16:32:22 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      278769       45396         368       61033      103489
Swap:          8191        5230        2961


Epoch 20 End 

16:33:16 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      283010       41155         368       61033       99248
Swap:          8191        5230        2961


Epoch 25 Beginning

16:34:33 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      284364       39797         368       61038       97894
Swap:          8191        5230        2961


Epoch 25 End 

16:39:46 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      287205       36955         368       61038       95053
Swap:          8191        5228        2963

 

Epoch 30 Beginning

16:41:14 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      289252       34906         368       61041       93008
Swap:          8191        5228        2963


Epoch 30 End 

16:47:35 [lsancere@sulaco TensorFlowGPU]# free -m
              total        used        free      shared  buff/cache   available
Mem:         385200      293019       31139         368       61041       89240
Swap:          8191        5228        2963

Regarding what you proposed, we add the following lines in our code to use a dynamic sequence:

ind = rng.permutation(len(Raw))
self.X= self.DataSequencer(Raw, self.axis_norm, Normalize = True, labelMe = False)
self.Y= self.DataSequencer(RealMask, self.axis_norm, Normalize = False, labelMe = True)

n_val = max(1, int(round(0.15 * len(ind))))
ind_train, ind_val = ind[:-n_val], ind[-n_val:]

self.X_val, self.Y_val = [self.X[i] for i in ind_val]  , [self.Y[i] for i in ind_val]
self.X_trn, self.Y_trn = [self.X[i] for i in ind_train], [self.Y[i] for i in ind_train]

With the following DataSequence class:

class DataSequencer(Sequence):
         
         
            def __init__(self, files, axis_norm, Normalize = True, labelMe = False):
                super().__init__() 
                
                self.files = files
               
                self.axis_norm = axis_norm
                self.labelMe = labelMe
                self.Normalize = Normalize
                
            def __len__(self):
                return len(self.files)
            
            
            def __getitem__(self, i):
                
                        #Read Raw images
                         if self.Normalize == True:
                                 x = ReadFloat(self.files[i]) 
                                 x = normalize(x,1,99.8,axis= self.axis_norm) 
                         if self.labelMe == True:
                                 #Read Label images
                                 x = ReadInt(self.files[i])
                                 x = label(x)
                   
                         return x

ReadFloat and ReadInt are the following:

def ReadFloat(fname):

    return imread(fname).astype('float16')         
         

def ReadInt(fname):

    return imread(fname).astype('uint16')  

The training images have size between 384 by 384 pixels to 4000 by 4000 pixels and we choose a patch size of 384 by 384 in out training with 128 Rays. Total number of Images = 4216

Do you have any clues about this memory increase?

Thank you for your support.

https://github.com/scijava/pom-scijava/blob/bed338cc73b46ddc104a4e871537803f700e2461/pom.xml#L419

These versions exist now!

Hey @bval, super sorry for never getting back to you. I spent a day trying to get the old bash script working again but would like to migrate it to AppleScript or Python. How do you imagine an Apple Script being able to import them into PrefPane? I'm not terribly familiar.

Also where is com.apple.TextInput? I don't see it under ~/Library/Preferences/