innat

import tensorflow as tf
from tensorflow.keras import layers

H_AXIS = -3
W_AXIS = -2

class RandomCutout(layers.Layer):
    """Randomly cut out rectangles from images and fill them.

    Args:

innat

Note

While writing a video data to tfrecord format, the output tfrecord file size would be much larger than the original video file. For quick demonstration purpose, some may use frame step to encode the frame to keep the overal size minimal. But in actual case (research or project) all frame should be considered while encoding to tfrecord. By doing so, while using the tfrecord in the training time, we can sample frames with different indices. Check this discussion. The following code is tested in tf 2.12.

video data layout

Let's say, we have a video data set in the following format.

innat

About: A simple demonstration to translate multihead self attention from PyTorch to Keras.

Multi-Head Self Attention

import torch
import torch.nn as nn

class TorchAttentionModel(nn.Module):

innat

# Ref: https://gist.github.com/Rocketknight1/efc47242914788def0144b341b1ad638

import math
import tensorflow as tf
from tensorflow.keras import layers

class TFAdaptiveAveragePooling(layers.Layer):
    def __init__(self, output_size, **kwargs):
        super().__init__(**kwargs)
        if not isinstance(output_size, (list, tuple)):

innat

from typing import Tuple

import tensorflow as tf
from keras import layers

def uniform_temporal_subsample(x, num_samples, temporal_dim=-4):
    """
    Uniformly subsamples num_samples indices from the temporal dimension of the video.
    When num_samples is larger than the size of temporal dimension of the video, it
    will sample frames based on nearest neighbor interpolation.

innat

# ref. https://www.tensorflow.org/tutorials/video/video_classification

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

class Conv2Plus1D(keras.Model):
    def __init__(self, filters, kernel_size, padding):
        """A sequence of convolutional layers 

innat

UNet With ImageNet as Backbone

from tensorflow import keras
from tensorflow.keras import layers as nn
from tensorflow.keras import backend as K
from tensorflow.keras import applications

def short_summary(model):
    trainable_count = np.sum(

innat

from keras_core import layers

def ResBlockIdentity(filter, stride):
    
    def apply(inputs):
        
        skip = inputs
        x = layers.Conv2D(filter, kernel_size=3, strides=stride, padding='same')(inputs)
        x = layers.BatchNormalization()(x)
        x = layers.Activation('relu')(x)

innat

SOURCE

https://brain2life.hashnode.dev/how-to-install-pyenv-python-version-manager-on-ubuntu-2004

Install all required prerequisite dependencies

sudo apt-get update; sudo apt-get install make build-essential libssl-dev zlib1g-dev \
libbz2-dev libreadline-dev libsqlite3-dev wget curl llvm \
libncursesw5-dev xz-utils tk-dev libxml2-dev libxmlsec1-dev libffi-dev liblzma-dev

innat

import tensorflow as tf
from tensorflow import keras 

class RGBShift(keras.layers.Layer):
    """RGBShift class randomly shift values for each channel of the input RGB image. 
    """
    def __init__(
        self,
        factor,
        seed=None,