QDeepLandia v1.0.0 #1
Conversation
|
Small comment about the versioning: instead of a |
There was a problem hiding this comment.
Fantastic work, thanks @SebastienPeillet !
I have some minor style comments, as well as a pair of remarks relative to the get_trained_model function, from the deeposlandia API, that could be updated here.
I point out some additional points, however we will address them in further PRs; I plan to write some related issues so as to be meticulous.
| locale_path = os.path.join( | ||
| os.path.dirname(__file__), | ||
| 'i18n', | ||
| 'thyrsis_{}.qm'.format(locale)) |
| self.initProcessing() | ||
|
|
||
| self.toolbar = QToolBar(tr("QDeepLandia_toolbar")) | ||
| self.toolbar.setObjectName("QDeepLandia_toolbar") |
There was a problem hiding this comment.
Not sure to understand the difference with the previous line, do you really want both these lines?
|
|
||
|
|
||
| class DatagenQDeepLandiaProcessingAlgorithm(QgsProcessingAlgorithm): | ||
| """ |
|
|
||
| path='' | ||
| for i in [dest_path, dataset, 'input','testing','images']: | ||
| path = os.path.join(path, i) |
There was a problem hiding this comment.
os.path does not seem very practical in this use case... Maybe we should consider pathlib.Path (Python3) in the future.
| shutil.rmtree(os.path.join(dest_path, dataset, 'preprocessed', str(shape))) | ||
|
|
||
| cmd = ['deepo', 'datagen', '-D', dataset, '-s', str(shape), '-P', dest_path, '-T', '1'] | ||
| subprocess.run(cmd) |
There was a problem hiding this comment.
I'm kind of puzzled with the call to subprocess. I'm sure we could use the deeposlandia API; let's say it's OK there, however it deserves a special issue.
| dataset = os.path.basename(os.path.abspath(os.path.join(os.path.dirname(model_path), '..', '..', '..'))) | ||
| image_size = os.path.splitext(os.path.basename(model_path))[0].split('-')[-1] | ||
| try : | ||
| model = get_trained_model(datapath, dataset, int(image_size), int(nb_labels)) |
There was a problem hiding this comment.
Same remark as in the datagen.py module.
delhomer
force-pushed
the
dev
branch
2 times, most recently
from
5806c54
to
0e5a51a
Compare
First operative version of qdeeplandia.
Long story short :
qdeeplandia.pycontains the plugin main class QDeeplandiaPlugin that :guifolder stores pop-up windows; so at the moment there is only the windows that ask the number of label handled by the loaded model.processing_provideris the folder relative to the processing toolbox, it contains :provider.py, the qdeeplandia processing tool providerdatagen.py, a processing tool to generate tiles according to the loaded model; for now it used a subprocessdeepo datagen etc etc, it could be changed to use API in the futureinference.py, a processing tool to do inference on the current raster layer, this tool use the previous one to generate the tiles.inferenceTask.py: Inference tool has been wrapped in a QgsTask to avoid to freeze QGIS during the inferenceInstallation
I didn't really care about the installation procedure. But to explain how it works for me :
futhermore there is a config.ini in the qdeeplandia repository, which is needed to initialize deeposlandia, but I believe that it is not used after initialization.