def test_non_native_dialog():
d = dict()
FileDialog._non_native_if_set(d)
assert "options" not in d
os.environ["USE_NON_NATIVE_FILE"] = "1"
FileDialog._non_native_if_set(d)
assert d["options"] == QtWidgets.QFileDialog.DontUseNativeDialog
def _add_model_action(self):
"""Method called when user clicks 'add models' button."""
dir = FileDialog.openDir(None, dir=None, caption="")
if dir:
self._cfg_getter.dir_paths.append(dir)
self._update_cfgs()
def visualizeOutputs(self):
"""Gui for adding overlay with live visualization of predictions."""
filters = ["Model (*.json)"]
# Default to opening from models directory from project
models_dir = None
if self.state["filename"] is not None:
models_dir = os.path.join(os.path.dirname(self.state["filename"]),
"models/")
# Show dialog
filename, selected_filter = FileDialog.open(
self,
dir=models_dir,
caption="Import model outputs...",
filter=";;".join(filters),
)
if len(filename) == 0:
return
# Model as overlay datasource
# This will show live inference results
from sleap.gui.overlays.base import DataOverlay
overlay = DataOverlay.from_model(filename,
self.state["video"],
player=self.player)
self.overlays["inference"] = overlay
self.plotFrame()
def select_file(*args, x=field):
filename = FileDialog.openDir(None,
directory=None,
caption="Open File")
if len(filename):
x.setText(filename)
self.valueChanged.emit()
def save(
self, output_dir: Optional[str] = None, labels_filename: Optional[str] = None
):
"""Save scripts and configs to run pipeline."""
if output_dir is None:
models_dir = os.path.join(os.path.dirname(self.labels_filename), "/models")
output_dir = FileDialog.openDir(
None, directory=models_dir, caption="Select directory to save scripts"
)
if not output_dir:
return
pipeline_form_data = self.pipeline_form_widget.get_form_data()
items_for_inference = self.get_items_for_inference(pipeline_form_data)
config_info_list = self.get_every_head_config_data(pipeline_form_data)
if labels_filename is None:
labels_filename = self.labels_filename
runners.write_pipeline_files(
output_dir=output_dir,
labels_filename=labels_filename,
config_info_list=config_info_list,
inference_params=pipeline_form_data,
items_for_inference=items_for_inference,
)
def test_non_native_dialog():
save_env_non_native = os.environ.get("USE_NON_NATIVE_FILE", None)
os.environ["USE_NON_NATIVE_FILE"] = ""
d = dict()
FileDialog._non_native_if_set(d)
assert "options" not in d
os.environ["USE_NON_NATIVE_FILE"] = "1"
FileDialog._non_native_if_set(d)
assert d["options"] == QtWidgets.QFileDialog.DontUseNativeDialog
if save_env_non_native is not None:
os.environ["USE_NON_NATIVE_FILE"] = save_env_non_native
def select_file(*args, x=field):
filter = item.get("filter", "Any File (*.*)")
filename, _ = FileDialog.open(
None, directory=None, caption="Open File", filter=filter
)
if len(filename):
x.setText(filename)
self.valueChanged.emit()
def doFileSelection(self):
"""Shows file browser to add training profile for given model type."""
filename, _ = FileDialog.open(
None,
dir=None,
caption="Select training configuration file...",
filter="JSON (*.json)",
)
return self._cfg_getter.try_loading_path(filename) if filename else None
def test_non_native_dialog():
save_env_non_native = os.environ.get("USE_NON_NATIVE_FILE", None)
os.environ["USE_NON_NATIVE_FILE"] = ""
d = dict()
FileDialog._non_native_if_set(d)
is_linux = sys.platform.startswith("linux")
if is_linux:
assert d["options"] == QtWidgets.QFileDialog.DontUseNativeDialog
else:
assert "options" not in d
os.environ["USE_NON_NATIVE_FILE"] = "1"
d = dict()
FileDialog._non_native_if_set(d)
assert d["options"] == QtWidgets.QFileDialog.DontUseNativeDialog
if save_env_non_native is not None:
os.environ["USE_NON_NATIVE_FILE"] = save_env_non_native
def visualizeOutputs(self):
"""Gui for adding overlay with live visualization of predictions."""
filters = ["Model (*.json)"]
# Default to opening from models directory from project
models_dir = None
if self.state["filename"] is not None:
models_dir = os.path.join(os.path.dirname(self.state["filename"]),
"models/")
# Show dialog
filename, selected_filter = FileDialog.open(
self,
dir=models_dir,
caption="Import model outputs...",
filter=";;".join(filters),
)
if len(filename) == 0:
return
# Model as overlay datasource
# This will show live inference results
from sleap.gui.overlays.base import DataOverlay
predictor = DataOverlay.make_predictor(filename)
show_pafs = False
# If multi-head model with both confmaps and pafs,
# ask user which to show.
if (predictor.confidence_maps_key_name
and predictor.part_affinity_fields_key_name):
results = FormBuilderModalDialog(
form_name="head_type_form").get_results()
show_pafs = "Part Affinity" in results["head_type"]
overlay = DataOverlay.from_predictor(
predictor=predictor,
video=self.state["video"],
player=self.player,
show_pafs=show_pafs,
)
self.overlays["inference"] = overlay
self.plotFrame()
def locateFile(self, idx: int):
"""Shows dialog for user to locate a specific missing file."""
old_filename = self.filenames[idx]
_, old_ext = os.path.splitext(old_filename)
caption = f"Please locate {old_filename}..."
filters = [f"Missing file type (*{old_ext})", "Any File (*.*)"]
new_filename, _ = FileDialog.open(None,
dir=None,
caption=caption,
filter=";;".join(filters))
if new_filename:
# Try using this change to find other missing files
self.setFilename(idx, new_filename)
# Redraw the table
self.file_table.reset()
def ask(self):
"""Runs the import UI.
1. Show file selection dialog.
2. Show import parameter dialog with widget for each file.
Args:
None.
Returns:
List with dict of the parameters for each file to import.
"""
filenames, filter = FileDialog.openMultiple(
None,
"Select videos to import...", # dialogue title
".", # initial path
"Any Video (*.h5 *.hd5v *.mp4 *.avi *.json);;HDF5 (*.h5 *.hd5v);;ImgStore (*.json);;Media Video (*.mp4 *.avi);;Any File (*.*)",
)
if len(filenames) > 0:
importer = ImportParamDialog(filenames)
importer.accepted.connect(lambda: importer.get_data(self.result))
importer.exec_()
return self.result
def export_package(self, output_path: Optional[str] = None, gui: bool = True):
"""Export training job package."""
# TODO: Warn if self.mode != "training"?
if output_path is None:
# Prompt for output path.
output_path, _ = FileDialog.save(
caption="Export Training Job Package...",
dir=f"{self.labels_filename}.training_job.zip",
filter="Training Job Package (*.zip)",
)
if len(output_path) == 0:
return
# Create temp dir before packaging.
tmp_dir = tempfile.TemporaryDirectory()
# Remove the temp dir when program exits in case something goes wrong.
# atexit.register(shutil.rmtree, tmp_dir.name, ignore_errors=True)
# Check if we need to include suggestions.
include_suggestions = False
items_for_inference = self.get_items_for_inference(
self.pipeline_form_widget.get_form_data()
)
for item in items_for_inference.items:
if (
isinstance(item, runners.DatasetItemForInference)
and item.frame_filter == "suggested"
):
include_suggestions = True
# Save dataset with images.
labels_pkg_filename = str(
Path(self.labels_filename).with_suffix(".pkg.slp").name
)
if gui:
ret = sleap.gui.commands.export_dataset_gui(
self.labels,
tmp_dir.name + "/" + labels_pkg_filename,
all_labeled=False,
suggested=include_suggestions,
)
if ret == "canceled":
# Quit if user canceled during export.
tmp_dir.cleanup()
return
else:
self.labels.save(
tmp_dir.name + "/" + labels_pkg_filename,
with_images=True,
embed_all_labeled=False,
embed_suggested=include_suggestions,
)
# Save config and scripts.
self.save(tmp_dir.name, labels_filename=labels_pkg_filename)
# Package everything.
shutil.make_archive(
base_name=str(Path(output_path).with_suffix("")),
format="zip",
root_dir=tmp_dir.name,
)
msg = f"Saved training job package to: {output_path}"
print(msg)
# Close training editor.
self.accept()
if gui:
msgBox = QtWidgets.QMessageBox(text="Created training job package.")
msgBox.setDetailedText("Path:\n" + output_path)
msgBox.setWindowTitle("Training Job Package")
okButton = msgBox.addButton(QtWidgets.QMessageBox.Ok)
openFolderButton = msgBox.addButton(
"Open containing folder", QtWidgets.QMessageBox.ActionRole
)
colabButton = msgBox.addButton(
"Go to Colab", QtWidgets.QMessageBox.ActionRole
)
msgBox.exec_()
if msgBox.clickedButton() == openFolderButton:
sleap.gui.commands.open_file(str(Path(output_path).resolve().parent))
elif msgBox.clickedButton() == colabButton:
# TODO: Update this to more workflow-tailored notebook.
sleap.gui.commands.copy_to_clipboard(output_path)
sleap.gui.commands.open_website(
"https://colab.research.google.com/github/murthylab/sleap-notebooks/blob/master/Training_and_inference_using_Google_Drive.ipynb"
)
tmp_dir.cleanup()