Installation instructions fail

[mastricker]

Hello, I read the corresponding publication to your code and landed here. In following your instruction for installation, I repeatedly arrive at the same error at the point of "check installation" with the command: python object_detection/builders/model_builder_tf2_test.py

class FreezableSyncBatchNorm(tf.keras.layers.experimental.SyncBatchNormalization
AttributeError: module 'keras._tf_keras.keras.layers' has no attribute 'experimental'

My original hunch was that the updated models repository does not play nicely with the rest of the explicit definitions of version numbers for dependencies, e.g. tensorflow version 2.7. I then did checkout the Dec 2022 version of the models repository -- with the same error.

Do you, by any chance, have an environment file for a working version of this repository?

Thank you, Markus

[ViktoriiaBaib]

Hello Markus,

Is next step python object_detection_axes_legend.py working for you?

Below is the content of the environment.yml file that works for me. Hope it helps.

name: tf2od
channels:
  - conda-forge
  - defaults
dependencies:
  - ca-certificates=2021.4.13=hecd8cb5_1
  - libcxx=10.0.0=1
  - libffi=3.3=hb1e8313_2
  - ncurses=6.2=h0a44026_1
  - openssl=1.1.1k=h9ed2024_0
  - pip=21.1.1=py38hecd8cb5_0
  - python=3.8.10=h88f2d9e_7
  - readline=8.1=h9ed2024_0
  - sqlite=3.35.4=hce871da_0
  - tk=8.6.10=hb0a8c7a_0
  - xz=5.2.5=h1de35cc_0
  - zlib=1.2.11=h1de35cc_3
  - pip:
      - absl-py==0.12.0
      - apache-beam==2.29.0
      - astunparse==1.6.3
      - attrs==21.2.0
      - avro-python3==1.9.2.1
      - cachetools==4.2.2
      - certifi==2020.12.5
      - cffi==1.14.5
      - chardet==4.0.0
      - contextlib2==0.6.0.post1
      - crcmod==1.7
      - cycler==0.10.0
      - cython==0.29.23
      - dill==0.3.1.1
      - dm-tree==0.1.6
      - docopt==0.6.2
      - easyocr==1.6.2
      - fastavro==1.4.1
      - flatbuffers==1.12
      - future==0.18.2
      - gast==0.4.0
      - gin-config==0.4.0
      - google-api-core==1.28.0
      - google-api-python-client==2.6.0
      - google-auth==1.30.1
      - google-auth-httplib2==0.1.0
      - google-auth-oauthlib==0.4.4
      - google-cloud-bigquery==2.17.0
      - google-cloud-core==1.6.0
      - google-crc32c==1.1.2
      - google-pasta==0.2.0
      - google-resumable-media==1.3.0
      - googleapis-common-protos==1.53.0
      - grpcio==1.34.1
      - h5py==3.1.0
      - hdfs==2.6.0
      - httplib2==0.17.4
      - idna==2.10
      - imageio==2.22.4
      - importlib-resources==5.1.4
      - joblib==1.0.1
      - kaggle==1.5.12
      - keras-nightly==2.5.0.dev2021032900
      - keras-preprocessing==1.1.2
      - kiwisolver==1.3.1
      - lvis==0.5.3
      - lxml==4.6.3
      - markdown==3.3.4
      - matplotlib==3.4.2
      - networkx==2.8.8
      - ninja==1.11.1
      - numpy==1.19.2
      - oauth2client==4.1.3
      - oauthlib==3.1.0
      - object-detection==0.1
      - opencv-python==4.5.2.52
      - opencv-python-headless==4.5.2.52
      - opt-einsum==3.3.0
      - packaging==20.9
      - pandas==1.2.4
      - pillow==9.3.0
      - portalocker==2.0.0
      - promise==2.3
      - proto-plus==1.18.1
      - protobuf==3.17.1
      - psutil==5.8.0
      - py-cpuinfo==8.0.0
      - pyarrow==3.0.0
      - pyasn1==0.4.8
      - pyasn1-modules==0.2.8
      - pyclipper==1.3.0.post4
      - pycocotools==2.0.0
      - pycparser==2.20
      - pydot==1.4.2
      - pymongo==3.11.4
      - pyparsing==2.4.7
      - python-bidi==0.4.2
      - python-dateutil==2.8.1
      - python-slugify==5.0.2
      - pytz==2021.1
      - pywavelets==1.4.1
      - pyyaml==5.4.1
      - requests==2.25.1
      - requests-oauthlib==1.3.0
      - rsa==4.7.2
      - sacrebleu==1.5.1
      - scikit-image==0.19.3
      - scikit-learn==0.24.2
      - scipy==1.4.1
      - sentencepiece==0.1.95
      - seqeval==1.2.2
      - setuptools==57.0.0
      - shapely==2.0.0
      - six==1.15.0
      - tensorboard==2.5.0
      - tensorboard-data-server==0.6.1
      - tensorboard-plugin-wit==1.8.0
      - tensorflow==2.5.0
      - tensorflow-addons==0.13.0
      - tensorflow-datasets==4.3.0
      - tensorflow-estimator==2.5.0
      - tensorflow-hub==0.12.0
      - tensorflow-metadata==1.0.0
      - tensorflow-model-optimization==0.5.0
      - termcolor==1.1.0
      - text-unidecode==1.3
      - tf-models-official==2.5.0
      - tf-slim==1.1.0
      - threadpoolctl==2.1.0
      - tifffile==2022.10.10
      - torch==1.13.0
      - torchvision==0.14.0
      - tqdm==4.61.0
      - typeguard==2.12.0
      - typing-extensions==3.7.4.3
      - uritemplate==3.0.1
      - urllib3==1.26.5
      - werkzeug==2.0.1
      - wheel==0.36.2
      - wrapt==1.12.1
      - zipp==3.4.1

[mastricker]

Hi @ViktoriiaBaib , thank you very much - I will try it and get back to you. Best, Markus

[mastricker]

Hi @ViktoriiaBaib , I think we are getting closer, but one issue still persits: The dependency - object-detection==0.1 needs to be installed from the cloned git repo. I checked out the Version from Dec 2022, the last date of the curedataextraction repository. But once I install, tensorflow will update to version 2.16 rather than stay on the version 2.5 as specified in your environment file.

The result are a lot of version incompatibility issue because the interfaces and general organization of the tensorflow library have changed significantly from version 2.5 to 2.16.

Would you be able to check which exact version (commit hash) from the models repository you checked out for the working version of curvedataextraction?

Thank, Markus

[qiao1119]

Hi @mastricker I have also encountered the same problem. Have you solved it Thank， Oranjia

[mastricker]

Dear @qiao1119 ,

we have not solved it yet. I have now asked a student to strategically go through the version of the models repository to find the one that presumably fits with the rest of the code. Will update as soon as I have news.

Cheers, Markus

[2DEG]

Hi @qiao1119, @mastricker and @ViktoriiaBaib

I managed to resolve the issues and successfully run the software! The main problem was with the models/research/object_detection itself due to outdated components.

1) Protobuf Version: Avoid the latest protoc. Instead, download the binary from this release and use it. In my case it looked like this:

   ~/Downloads/protoc-22.5-linux-x86_64/bin/protoc object_detection/protos/*.proto --python_out=.

2) Package Versions: I settled on using protobuf v3.20.3, keras v.2.13.1, and tensorflow v.2.13.0 installed via pip. Despite some initial compatibility warnings with other packages, these did not affect the operation. The system warned about multiple conflicts with tensorflow-text, tf-keras, tf-models-official, and torch, but I was able to proceed without any hindrance.

3) SciPy Issue: There was another hiccup at step 19 related to a SciPy error:

   TypeError: Argument `a` is not recognized as numeric.

A simple solution here is to downgrade SciPy or update the problematic code segment.

After addressing these points, everything worked smoothly!

Best, Nick