tensorflow tensorflow2.0 object-detection-api

Training modern object detection with Tensorflow

The Tensorflow object detection API is deprecated, and refers to Tensorflow-vision. Unfortunately, the documentation, as far as I can see, is just links to a series of preprints. All the other references go back to the deprecated API.

So, how can I train an object detection model with Tensorflow in 2023?

Solution

It seems the functionality has been moved to Keras CV. The examples are here.

TensorFlow GPU problem 'libnvinfer.so.7' and ' 'libnvinfer.so.7'' could not load
How to extract and save images from tensorboard event summary?
Layer 'conv2d_11' expected 2 variables, but received 0 variables during loading. Expected: ['conv2d_11/kernel:0', 'conv2d_11/bias:0']
Division by neural network; fixed value produced each time
Cuda 12 + tf-nightly 2.12: Could not find cuda drivers on your machine, GPU will not be used, while every checking is fine and in torch it works
AssertionError: Some objects had attributes which were not restored
ImportError : import segmentation_models as sm
tensorflow:Can save best model only with val_acc available, skipping
How to choose the number of hidden layers and nodes?
TensorFlow libdevice not found. Why is it not found in the searched path?
Issue with training Keras model using ModelCheckpoint in Kaggle notebook (Unexpected result of `train_function` (Empty logs))
KeyError: 'filename' (Pandas)
What is the difference between backpropagation and reverse-mode autodiff?
tensorflow - how to use 16 bit precision float
Memory reduction Tensorflow TPU v2/v3 bfloat16
tensorflow: register numpy bfloat16 extension
Nvidia Cudatoolkit vs Conda Cudatoolkit
ImportError: No module named tensorflow.keras.applications
Huggingface - Finetuning in Tensorflow with custom datasets
Why is the accuracy for my Keras model always 0 when training?
Importing pandas before tensorflow makes the script freeze
tf.function converts variable to tensor automatically
AttributeError: module 'tensorflow' has no attribute 'compat' when loading tf.compat.v1.train.SessionRunHook
Keras model.export() fails because of NoneType shapes in model
Getting error:([WinError 123] The filename, directory name, or volume label syntax is incorrect: '//') while training custom dataset through mask rcnn
Single-headed transformer model, output and target tensor sizes don't match
module 'keras.utils.generic_utils' has no attribute 'get_custom_objects' when importing segmentation_models
Inconsistency in input shape to neural network
ImportError: cannot import name 'model_lib_v2' from 'object_detection' (already installed an still not working)
AttributeError: The layer has never been called and thus has no defined input shape