Recently Keras & Tensorflow are the most used advanced Machine Learning and Deep Learning Libraries in the market, with the development in the application of AI and ML many companies are implementing Keras models in their systems and business strategies, Here are some of the example of Keras or Keras Example that we can make to get know to this highly advanced field of the future computer sector.
- What is Keras:
- Why we learn Keras
- Practical Keras Examples with Applications:
- 1) Keras in Rock Paper Scissor classification:
- 2) Keras in Image Classification with the help of OpenCV:
- 3) Keras in Computer Vision for Facial Expressions Recognition:
- 4) Keras in Face Mask Detection with help of OpenCV & Tensorflow:
- 5) Keras in Training a Generative Adversarial Networks (GAN):
- 6) Keras in Heart Disease Classification:
- 7) Example of Keras in MobileNet Image Classifier:
- 8) Cat vs Dog classification in Keras
What is Keras:
Keras is a powerful, easy-to-use open-source library for the development of the artificial neural network, and deep-learning models Keras is also known as python’s deep learning API.
Keras also has support for convolutional and recurrent neural networks, which helps in the development of dynamic deep learning models, with the help of Tensorflow libraries.
The main purpose of Keras is to train user-friendly Deep-learning models. it is also known as high-level neural network API which python is the main programming language and has the ability to support multiple back-end neural engines.
Why we learn Keras
Keras is known for its user-friendliness, apart from that it requires python as the main programming language which is one of the best programming language to learn in current market.
Keras is also modular and optimistically easy to extend as well, its well-developed API is quite helpful in following best practices and reducing complex cognitive load.
There are many deep learning modules that are helpful in the development of new user-friendly modules like, Neural layers, optimizers, activation functions, cost functions, optimizers, initialization schemes, regularization schemes, etc are some of the stand-alone modules that we can combine to create new models from it.
Practical Keras Examples with Applications:
Keras is an open-source deep-learning and neural-net library that acts as an interface to Tensorflow Library, here are some of the Examples of Keras where it can be applied.
1) Keras in Rock Paper Scissor classification:
As we know games are the best way to increase cognitive thinking ability, likewise, Deep learning and Neural Networks can be better trained by simulating games, here is an example of the game known as Rock-Paper-Scissor.
In the previous blog, we had an example of Reinforcement learning playing the game of pong, in that we trained the Reinforcement learning model how to play the game of pong by itself and excel at it, for more visit “TOP 19 PRACTICAL REINFORCEMENT LEARNING EXAMPLES”
Here we are taking the Keras example of making a self-playing game of Rock-Paper-Scissors, in this machine-learning model in this convolution Neural Network is used to create hand gesture images of Rock, Paper, and Scissor.
In this module the AI uses your laptop’s webcam to play with another player or with an opponent we can say, to make this self-playing game we have to use a pre-trained convolution neural net known as Squeeze Net and retrain the output layer as three categories (rock, paper, scissor).
The following image demonstrates the steps required to make a game of Rock, paper, scissors in keras.
We can train our own data set by using our webcam, thereby we can save time as well.
Here are the final results of the example of the keras working model.
The Github link for the following project is rock-paper-scissors
2) Keras in Image Classification with the help of OpenCV:
Image classification is one of the most practiced projects in machine learning, deep learning, & neural networks as well. there are many uses of applying image classification right from gender detection to security surveillance checkups, with the help of Keras and TensorFlow libraries we can easily implement image classification modules in our projects.
The following video is the best tutorial for beginners to start image classification with Keras.
The reason why learning image classification is important because it applies in every field that is related to computer vision, image mapping, data classification, and much more.
Here is the official repository of Keras projects “Image classification from scratch”
3) Keras in Computer Vision for Facial Expressions Recognition:
There are many face related projects in Keras, this is one of them in this Keras example we are taking multiple datasets of different facial expression images, these images are then fed to the Keras, OpenCV, Tensorflow models.
Here is the end result of the Keras project,
Here is the google drive link for downloading Datasets face-expression-recognition-dataset
Here is the Github link “Facial-Expressions-Recognition”
4) Keras in Face Mask Detection with help of OpenCV & Tensorflow:
Recently many YouTubers and students are trying to implement facemask detection in Keras, OpenCV & Tensorflow, Due to pandemic many people are not wearing a mask, Hence to tackle the situation many public places have made compulsory to wear a mask.
But the problem is we cannot place a human to check whether the person is wearing a mask or not, it will be too stressful and risky for the person as well.
For that purpose, we are creating a Keras & OpenCV model that detects whether the person wearing a mask or not, I this model we are feeding a huge dataset of images of the person wearing a mask and a person not wearing a mask.
The following image represents the roadmap for this project.
Here’s the training accuracy of the Keras example project
Here’s the Github link “Face-Mask-Detection“
5) Keras in Training a Generative Adversarial Networks (GAN):
Generative Adversarial Networks is a deep learning method that is used to new output from older datasets that were provided by us to train itself, Generative Adversarial Networks comes under Machine learning modulation, in the subcategory of supervised machine learning also known as “Generative modeling”.
GANs is used to generate the newer type of images by computing previous images and their specialties using randomization, the specialty of GANs is that it has the ability to model high-dimensional data, Image super-resolution, creating new art, to provide multiple possible answers to multiple values.
As we discussed earlier GANs are based on unsupervised learning, here are some of the tree diagrams of supervised & unsupervised learning.
Both supervised and unsupervised is followed by Discriminative Modeling & Generative Modeling
For more have a look at “Overview of GAN Structure”
Here are some demo of GANs accuracy modelling.
The following image is an Example in Keras which shows constant & drastic improvement in image distillation in a lifespan of 4 years.
Here’s the Github link “t81_558_deep_learning“
6) Keras in Heart Disease Classification:
There are many fields where Artificial Intelligence and Machine Learning are applied extensively and proven to be quite beneficial as well, specifically in the field of Healthcare.
The application of Machine learning in the Healthcare industry is dynamically booming due to its accuracy and errorless implementation and also in data modeling. Here’s another application in Keras i.e is heart disease classification.
Heart Disease Classification works on multiple parameters, like previous operations, other diseases, genetic disorders, eating habits, etc. We can find a vast variety of dataset on Kaggle which is named “Heart Disease” it’s a Machine Learning Repository.
In Kaggle there’s an in-depth article on the implementation of Keras in heart disease classification, check here “Classification of Patient Heart Disease with Keras – MLP”
Here’s another video tutorial as well,
The Github Link for this project is “Predicting-Heart-Disease“
7) Example of Keras in MobileNet Image Classifier:
MobileNet a class of lightweight deep convolutional neural networks that are much smaller and faster in size than many of the mainstream popular models that are really well-known.
MobileNets are a class of small low-power low latency models that can be used for things like classification detection and other things that convolutional neural networks are typically good for and because of their small size these models are considered great for mobile devices hence the name mobile nets.
So I have some stats taken down here so just to give a quick comparison in regards to the size of the full VG g16 network that we’ve worked within the past, is about five hundred and fifty-three megabytes on disk so pretty large generally speaking the size of one of the currently largest mobile Nets is only about 17 megabytes so that’s a pretty huge difference especially when you think about deploying a model to run on a mobile app, for example, this vast size difference is due to the number of parameters or weights and biases contained in the model.
Here’s the youtube tutorial “MobileNet Image Classification with TensorFlow’s Keras API”
8) Cat vs Dog classification in Keras
It’s just a simple project, not that complicated as you think, basically, it uses Keras API to classify dogs and cats with help of image classification neural networks.
Here’s the download link for the cat and dog dataset “Kaggle Cats and Dogs Dataset”
The link to the Github repository “Cats-vs-Dogs-CNN-using-Keras“
Youtube link for tutorials “Cats vs Dogs Classifier using Keras- Live Project”