Read-only file system error

[Nicolas-Isaza]

Hello,

I've been using the ensemble over the last few weeks with no issues but started running into the below error sometime in the last few days. I am running the same setup as before (python -m speciesnet.scripts.run_model --folders "c:\your\image\folder" --predictions_json "c:\your\output\file.json") in Colab. Any insight on this would be great appreciated!

2025-04-11 12:52:10.659618: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in _run_code
File "/usr/local/lib/python3.11/dist-packages/speciesnet/scripts/run_model.py", line 456, in
app.run(main)
File "/usr/local/lib/python3.11/dist-packages/absl/app.py", line 308, in run
_run_main(main, args)
File "/usr/local/lib/python3.11/dist-packages/absl/app.py", line 254, in _run_main
sys.exit(main(argv))
^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/speciesnet/scripts/run_model.py", line 378, in main
model = SpeciesNet(
^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/speciesnet/multiprocessing.py", line 606, in init
self.classifier = SpeciesNetClassifier(
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/speciesnet/classifier.py", line 66, in init
self.model_info = ModelInfo(model_name)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/speciesnet/utils.py", line 116, in init
with open(filepath, mode="wb") as fp:
^^^^^^^^^^^^^^^^^^^^^^^^^
OSError: [Errno 30] Read-only file system: '/kaggle/input/speciesnet/keras/v4.0.0a/3/https___github.com_agentmorris_MegaDetector_releases_download_v5.0_md_v5a.0.0.pt'

[agentmorris]

Sorry you're having trouble with SpeciesNet. I am able to reproduce this issue on Colab; there appears to be an unexpected interaction between the model download system and the way Colab handles files. We will leave this issue open until we make a fully-functional Colab notebook, but the following workaround is working for me... in your Colab, force the model to get downloaded to a particular folder, like this:

import kagglehub
import os

os.environ['KAGGLE_CONFIG_DIR'] = '/content/models' 
model_path = kagglehub.model_download("google/speciesnet/keras/v4.0.0a",
                                force_download=True)

Then when it's time to run your model, tell it to read the model from that folder using the "--model" options, like this... i.e., instead of this:

!python -m speciesnet.scripts.run_model --folders "/content/images" --predictions_json "/content/images/stuff.json"

...do this:

!python -m speciesnet.scripts.run_model --folders "/content/images" --predictions_json "/content/images/stuff.json" --model $model_path

Let us know whether that works for you?

[Nicolas-Isaza]

Appreciate the quick reply. I tried your workaround but it doesn't seem to be working on my end:

!pip install speciesnet

import kagglehub
import os

os.environ['KAGGLE_CONFIG_DIR'] = '/content/models' 
model_path = kagglehub.model_download("google/speciesnet/keras/v4.0.0a",
                                force_download=True)
!python -m speciesnet.scripts.run_model --folders "/content/images" --predictions_json "/content/images/stuff.json" --model $model_path

2025-04-11 22:14:47.074964: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/usr/local/lib/python3.11/dist-packages/speciesnet/scripts/run_model.py", line 456, in <module>
    app.run(main)
  File "/usr/local/lib/python3.11/dist-packages/absl/app.py", line 308, in run
    _run_main(main, args)
  File "/usr/local/lib/python3.11/dist-packages/absl/app.py", line 254, in _run_main
    sys.exit(main(argv))
             ^^^^^^^^^^
  File "/usr/local/lib/python3.11/dist-packages/speciesnet/scripts/run_model.py", line 378, in main
    model = SpeciesNet(
            ^^^^^^^^^^^
  File "/usr/local/lib/python3.11/dist-packages/speciesnet/multiprocessing.py", line 606, in __init__
    self.classifier = SpeciesNetClassifier(
                      ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/dist-packages/speciesnet/classifier.py", line 66, in __init__
    self.model_info = ModelInfo(model_name)
                      ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/dist-packages/speciesnet/utils.py", line 116, in __init__
    with open(filepath, mode="wb") as fp:
         ^^^^^^^^^^^^^^^^^^^^^^^^^
OSError: [Errno 30] Read-only file system: '/kaggle/input/speciesnet/keras/v4.0.0a/3/https___github.com_agentmorris_MegaDetector_releases_download_v5.0_md_v5a.0.0.pt'

[agentmorris]

Sorry that didn't work, I don't have a good explanation for why this worked for me but not for you. I've put together a self-contained Colab notebook that handles this issue slightly differently, and I'm able to run the whole thing on Colab. This notebook:

• Installs the speciesnet package
• Downloads the model
• Copies it to a writeable folder before using it
• Downloads two test images from this repo
• Processes them using both the CLI (i.e., python -m speciesnet.scripts.run_model) and Python code

As is often the case when you !pip install a library on Colab, you will likely have to run the first code cell (which just does "!pip install speciesnet"), then re-start your Colab runtime, then you should be able to run the whole notebook.

Confirm that notebook works for you before making any changes?

[Nicolas-Isaza]

Thanks! That notebook runs fine on my end.