Sebastian Szymański

Hi, sorry to bother but I'm curious whether you will be releasing EfficientNetV2-lite model variants (as similar to their V1 counterparts)? Best regards, Sebastian

To reproduce: 1. git clone, cd and Run docker: `docker run -it --rm --gpus all -w /workspace/ -v ${PWD}:/workspace/ tensorflow/tensorflow:custom-op-gpu-ubuntu16` 2. Attempt to build the op: `make time_two_op` Will result...

Redundant bias paramter in ResNet's conv2d.

1

**Short description** In `tf.keras` by default Conv2D has bias enabled. We should explicitly set `use_bias=False` if the convolution layer is followed by BatchNormalization. This is the done in some existing...

Add EfficientNet Lite model variants.

2

**Short Description** EfficientNet Lite models are lighter variants of EfficientNets more suitable for mobile deployment. **Existing Implementations** Original implementation - [link](https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet/lite) My reimplementation - [link](https://github.com/sebastian-sz/efficientnet-lite-keras) **Other Information** EfficientNet Lite's have...

Added EfficientNet lite option.

4

# What does this PR do This PR adds `lite` option to existing EfficientNet V1's. EfficientNet lite's are slightly modified efficientnets that better support edge devices. The models and weights...

# What does this PR do? The `correct_pad` function in efficientnet v1 is using the wrong `image_dim`. It assumes `channels_first`, while keras-cv operates on `channels_last`. Reference function: [here](https://github.com/keras-team/keras/blob/bb78a0caaea0503728784c0fa5cf547c623633b0/keras/applications/imagenet_utils.py#L441) The weights...

630/add efficientnet lite

1

# What does this PR do? Adds EfficientNet Lite variants to `keras_cv` models. Fixes #630 This is a port of PR from the [Keras repository](https://github.com/keras-team/keras/pull/16905), as per [this comment].(https://github.com/keras-team/keras/pull/16905#issuecomment-1262811641) ##...

`vectorized_map` causes `tf.function` retracing.

20

### Problem description It seems like applying some layers that use `BaseImageAugmentationLayer` and `self.auto_vectorize=True`, over batched input are causing `tf.function` retracing: ```python layer = Solarization() # or Equaliztion() rng =...

Vectorize (+ randomize) Posterization.

1

**Short Description** Posterization could also benefit from vectorization and using factor sampler. This would not change performance much, but could allow to apply different augmentations per sample (currently this is...