[Python] Unable to bundle pyqt5 app with tensorflow_hub

I have a pyqt5 GUI app (Selecta) which runs perfectly if I run it as a script Selecta.py or if I compile it with the following spec file:

# -*- mode: python ; coding: utf-8 -*-
# from PyInstaller.utils.hooks import collect_data_files

a = Analysis(
['Selecta.py'],
pathex=[],
binaries=[],
datas=[
("src/selecta/yamnet-tensorflow2-yamnet-v1", "yamnet-tensorflow2-yamnet-v1"),
],
hiddenimports=[
"scipy._cyutility",
],
excludes=[],
hookspath=[],
hooksconfig={},
runtime_hooks=[],
noarchive=False,
optimize=0,
)
pyz = PYZ(a.pure)

exe = EXE(
pyz,
a.scripts,
[],
exclude_binaries=True,
name='Selecta',
debug=False,
bootloader_ignore_signals=False,
strip=False,
upx=True,
console=False,
disable_windowed_traceback=False,
argv_emulation=False,
target_arch=None,
codesign_identity=None,
entitlements_file=None,
)
coll = COLLECT(
exe,
a.binaries,
a.datas,
strip=False,
upx=True,
upx_exclude=[],
name='Selecta',
)
app = BUNDLE(
coll,
name='Selecta',
icon="icon.icns",
bundle_identifier=None,
)


I then tried to compile it into a .dmg using create-dmg. However when I run the .dmg, copy the app to my /Applications dir and open it I get an error:

Failed to execute script 'Selecta' due to unhandled exception: Op with name ConstructionFails


I think the error is related to how pyinstaller packages up tensorflow but I'm unsure how to address this. I just want my app to work as an Application the same way it runs as a script.

Happy to provide further context if needed.

Continue reading...