TensorFlow Official Models 2.13.1

This release fix breakage in TF models official pip package by restricting the PyYAML version to <5.4.0 and uses latest version of Cython.

TensorFlow Official Models 2.12.1

This release fixes the issue of preemption_watcher which would cause failed cases in TPU environments.

TensorFlow Official Models 2.13.0

This release of the Official Models targets TensorFlow 2.13.0. Note that Research/tutorial/sample models have been removed.

New features:

1. Checkpoints added for ResNet-(50, 101, 152), SpineNet-(49, 96, 143) and DeepLabV3+.
2. Implementation of Perceiver IO: A General Architecture for Structured Inputs & Outputs.
3. Implementation of Linformer: Self-Attention with Linear Complexity with tasks on long range arena.

Release branch is: https://github.com/tensorflow/models/tree/r2.13.0

TensorFlow Official Models 2.11.6

This release of the Official Models targets TensorFlow 2.11.6. Note that Research/tutorial/sample models have been removed.

TensorFlow Official Models 2.12.0

This release of the Official Models targets TensorFlow 2.12.0. Note that Research/tutorial/sample models have been removed.

TensorFlow Official Models 2.11.5

This release fixes various typing annotation issues in official/vision that do not work for Python version <3.9.

TensorFlow Official Models 2.11.4

This release includes a few updates to RetinaNet model under official/vision:

1. Skip generating image_info in SavedModel is input_type is tflite.
2. Normalize anchors during TFLite post-processing in object detection.
3. Add v3 NMS for Edge TPU.
4. Support class agnostic NMS for v2 NMS.
5. Support output decoded boxes (before NMS) even when NMS is applied.
6. A few other bug fix to make detection generator more robust.

TensorFlow Official Models 2.11.3

Fix typing annotation in vision model export that breaks package import.

TensorFlow Official Models 2.11.2

Match the changes with Pypi package and tensorflow.org documentation.

TensorFlow Official Models 2.11.0

This release of the Official Models targets TensorFlow 2.11.0. Note that Research/tutorial/sample models have been removed.

New features:

1. MobileNetV2 backbone for Mask RCNN
2. Panoptic deeplab supported;
3. freeze_backbone supported, and many bug fixes
4. FNet and SparseMixer implementations
5. Move ViT models from projects to vision main backbone folder

Release branch is: https://github.com/tensorflow/models/tree/r2.11