2️⃣ Understand the dataset

   The dataset which was used is a subset of the dataset from(https://www.kaggle.com/datasets/dheerajperumandla/drowsiness-dataset)
   it has 4 folder which are
   1) Closed_eyes - having 726 pictures
   2) Open_eyes - having 726 pictures
   3) Yawn - having 725 pictures
   4) no_yawn - having 723 pictures

3️⃣ Data preprocessing

1. preprocess the images from the closed_eye, open_eye, yawn and no_yawn folder.

2. Resizing all the images to the same dimensions and converting the images into numpy arrays.

3. The dataset will be split into training, validation, and testing sets.

4️⃣ Build and train the CNN model

The CNN model is designed and trained to classify images as either the driver's eye is opened or closed.

5️⃣ Train the model using Intel OneAPI to get better results

How does OneApi provide better performance :

Today’s computer systems are heterogeneous and include CPUs, GPUs, FPGAs, and other accelerators. The different architectures exhibit varied characteristics that can be matched to specific workloads for the best performance. Having multiple types of compute architectures leads to different programming and optimization needs. oneAPI and SYCL provide a programming model, whether through direct programming or libraries, that can be utilized to develop software tailored to each of the architectures.

Advantages of using OneAPI :

1. We can use Single code for both CPU and GPU (heterogeneous computing)
2. To implement machine learning based IoT projects easily with less hardwares as the machine learning part happens in cloud
3. To process files faster ie. it takes less time to run the epochs
4. OneAPI allows users to transcend Hardware restrictions and provide better performance for low powered computers
5. Accuracy will improve while using OneAPI

To migrate your project to OneAPI : click here! to get started

For reference : click here!

6️⃣ Save the model

   save the model to calculate the accuracy and loss

Accuracy and Loss

We did 80 epochs, to get a good accuracy from the model i.e. 98% for training accuracy and 97% for validation accuracy.

Output

Eyes opened 👀

Eyes closed

Learnings

1. Building a CNN model using Intel oneDNN : OneAPI is an open-source software toolkit developed by Intel that simplifies the development of high-performance, heterogeneous applications. It allows developers to write code that can run efficiently on a variety of architectures, including CPUs, GPUs, and FPGAs. oneDNN (Deep Neural Network) is a part of oneAPI and is an optimized library for deep learning. It provides highly optimized building blocks for neural network models that run efficiently on a variety of hardware platforms.

2. Machine Learning : How to use machine learning for identifiying the facial features from a drivers face to detect drowsiness.

3. Convolutional Neural Network(CNN) : How to build, train and fine-tune convolutional neural networks for image and video classification.

4. Preprocessinig the datasets : How to preprocess the data dowloaded from kaggle so that the machine learning could happen in a much better and efficient way.

5. Different aspects of drowsiness during driving : I have studied the different causes and reasons for drowsiness to occur and how to resolve it.

6. Facial regions and drowsiness: I have studied how the facial landmarks are identified using an ocular feature called eye aspect ratio and face landmarks.Eye regions are detected and extracted as regions of interest using a facial cue detector and The eye aspect ratio values are then calculated, analyzed and saved for each image.

7. Team work : Collaborating and communicating effectively in a team to deliver a project.

8. Understanding the need for a drowsiness detection system in vehicles These are just a few examples of the knowledge and skills that i likely gained while building this project. Overall, building a drowsiness detection model is a challenging and rewarding experience that requires a combination of technical expertise and knowledge on facial cue detection.

Project Deployment

We have built an app using Flutter. Flutter helps Build, test, and deploy beautiful mobile, web, desktop, and embedded apps from a single codebase. It is a cross-platform app development framework by Google which goes hand in hand with the model to help ensure the safety of the user and other commuters.

As soon as the model detects drowsiness, the model will send an API request call to the client app, which notifies the user to take some rest an