TensorFlow Object Detection API Installation with Specific TensorFlow Version

General Discussion
#1

Hello,

Is it possible to install the TensorFlow Object Detection API with a specific version (TF2.3)? I have TensorFlow 2.3 installed and I need to be using the API with that specific version, but when I try to install it it automatically uninstalls TF2.3 and installs TF2.5.

Any advice?

#3

You can clone repo with 2.3 tag.

#4

How do I pass the tag exactly?

#6

You can download object detection repo with specific tag - 2.3.0. Then install it with setup.py file.

#7

The link to this repo says that TensorFlow 2.x is not supported. It seems like it only supports TensorFlow 1.x.

#8

Hi There! sorry for reviving this post, but im in a similar issue where i need to install object detection API with tensorflow 2.10.0, but it forces me to install 2.11.0

So as you suggested i tried to clone the specific tag so i can avoid that problem:
git clone -b v2.10.0 GitHub - tensorflow/models: Models and examples built with TensorFlow

but! that tag (and many other tags n branches) does not contain the “research” folder, where all of the detection API resides.

So what am i missing?

Best!

1 Like
#10

You have to make a small correction in the “models/research/object_detection/packages/tf2/setup.py”…

“”“Setup script for object_detection with TF2.0.”""
import os
from setuptools import find_packages
from setuptools import setup

REQUIRED_PACKAGES = [
# Required for apache-beam with PY3
‘avro-python3’,
‘apache-beam’,
‘pillow’,
‘lxml’,
‘matplotlib’,
‘Cython’,
‘contextlib2’,
‘tf-slim’,
‘six’,
‘pycocotools’,
‘lvis’,
‘scipy’,
‘pandas’,
‘tf-models-official>=2.5.1’,
‘tensorflow_io’,
‘keras’
]

setup(
name=‘object_detection’,
version=‘0.1’,
install_requires=REQUIRED_PACKAGES,
include_package_data=True,
packages=(
[p for p in find_packages() if p.startswith(‘object_detection’)] +
find_packages(where=os.path.join(’.’, ‘slim’))),
package_dir={
‘datasets’: os.path.join(‘slim’, ‘datasets’),
‘nets’: os.path.join(‘slim’, ‘nets’),
‘preprocessing’: os.path.join(‘slim’, ‘preprocessing’),
‘deployment’: os.path.join(‘slim’, ‘deployment’),
‘scripts’: os.path.join(‘slim’, ‘scripts’),
},
description=‘Tensorflow Object Detection Library’,
python_requires=’>3.6’,
)

Line-No:21 ‘tf-models-official>=2.5.1’,
change is to the tf version which you have already installed…
If your tf version installed is 2.9.1, then this line has to be changed as below.,
‘tf-models-official==2.9.1’
This will ensure, while the code runs to install the OD API, it does not replace the tf-version already installed…

Regards…Kannan Rama