NTI_AI_Project_graphic era hill_2022.pptx

Creating a 3,000-word description of a Git assignment in PDF format involves detailing the assignment's objectives, instructions, use cases, and step-by-step procedures. Below is an outline for the description:

Git Assignment Overview
1. Introduction
What is Git?

Definition and significance of Git in software development.
Overview of version control systems and how Git revolutionized collaborative work.
Purpose of the Assignment:

The objective is to familiarize students with Git's basic and advanced features.
Importance of understanding Git for effective version control and collaboration in software projects.
2. Learning Outcomes
By the end of this assignment, students should be able to:
Initialize a Git repository.
Track changes and manage versions of files using Git.
Understand and implement branching and merging strategies.
Use Git commands for collaboration, such as pull, push, clone, and fetch.
Resolve conflicts in code versions.
Use Git for continuous integration and deployment (optional advanced section).
3. Assignment Tasks
Task 1: Git Setup and Configuration

Install Git on your local machine.
Configure Git with your username and email.
Initialize a new Git repository.
Task 2: Basic Git Commands

Create a new directory and initialize it as a Git repository.
Create a new file and make initial commits.
View the commit history using git log.
Explore git status and git diff commands to track changes.
Task 3: Working with Branches

Create a new branch from the main branch.
Switch between branches and understand the purpose of branching.
Merge changes from one branch to another and handle merge conflicts.
Task 4: Collaborating with Git

Clone an existing repository from a remote server (e.g., GitHub).
Make changes, commit them, and push to the remote repository.
Use pull requests to propose changes and review code collaboratively.
Task 5: Advanced Git Features (Optional)

Explore rebasing vs. merging.
Stash changes and apply them later.
Use Git hooks for automating tasks.
Explore Git tagging and release management.
4. Detailed Step-by-Step Instructions
Step 1: Setting Up Git

Download and install Git from the official website.
Configure global user settings (git config --global user.name "Your Name", git config --global user.email "[email protected]").
Step 2: Initializing a Repository

Create a directory for your project.
Initialize Git with git init.
Add files to the staging area with git add.
Commit changes with git commit -m "Initial commit".
Step 3: Branching and Merging

Create a new branch with git branch .
Switch to the branch with git checkout .
Merge changes with git merge .
Resolve any conflicts that arise during merging.
Step 4: Working with Remote Repositories

Clone a repository using git clone .
Push changes to a remote repository using git push.
Pull updates from a remote repository with git pull.
Step 5: Advanced Features

1. CNN                               Doctork Application

2. INTROUDUCTION ABOUT National Telecommunications Institute Artificial Intelligence Summer Training 2022 Artificial Intelligence Course Final Project Under Supervision of : Engineer Eman Negm Students:          1)Ahmed Mohamed          2)Adham Khaled          3)Sarah Hesham          4)Reem Ali          5)Doaa Salah Faculty of Computers and Artificial Intelligence (Bioinformatics Department & Mainstream) Faculty of Engineering (Biomedical Engineering) Cairo University & Helwan University

3. Main Idea A health application system with simple GUI Pyqt5 used for GUI Python Programming Language Used Deep Learning Models used Covid19_Pneumonia_Normal Chest X-rays detection Breast Cancer detection Brain Tumors detection You simply browse and upload your image and click a button to detect and classify the image  H5 files deployed and trained Accuracies >92% CNN Models

4. What is Doctork?? Doctork is a Simple Desktop Application That does a prediction of several disease from rays It can predict: Covid19 & Pneumonia Breast Cancer Brain Cancer

5. How we did it?

6. First:  Brief About CNN What is the meaning of CNN Usage Layers in a Convolutional Neural Network

7. 1:CNN meaning CNN refer to   Convolutional Neural Network. type of artificial neural network, which is widely used for image/object recognition and classification. Deep Learning thus recognizes objects in an image by using a CNN.

8.   2:Usage                               Image recognition CNNs are often used in image recognition systems Video analysis Compared to image data domains, there is relatively little work on applying CNNs to video classification  Anomaly Detection A CNN with 1-D convolutions was used on time series in the frequency domain by an unsupervised model to detect anomalies in the time domain

9. Drug discovery By Predicting the interaction between molecules and biological proteins Health risk assessment and biomarkers of aging discovery Video analysis Some extensions of CNNs into the video domain have been explored

10.    3: Layers in a Convolutional Neural Network A convolution neural network has multiple hidden layers that help in extracting information from an image:       1:  Convolution layer       2:Pooling layer       3:Fully connected layer      

11.                                                       Convolution layer It is the first step to extract features from image  It has several filters that perform the convolution operation. Kernel is a filter  kernel is  a matrix that moves over the input data , performs the dot product with the sub-region of input data, and gets the output as the matrix of dot products. Multiple kernels work as different feature extractors, such as a horizontal edge detector, a vertical edge detector,  and an outline detector.  

12. Calculating the Output Dimension n: the size of input image f: filter size p: padding( Valid,Same ) s: stride

13. And use activation function such Relu or Tanh

14. Pooling Layer The pooling layer replaces the output of the network at certain locations by deriving a summary statistic of the nearby outputs helps in reducing the spatial size of the representation, which decreases the required amount of computation and weights processed on every slice of the representation individually.

15.                                      Types of Pooling Layer Max Pooling It is a pooling operation that selects the maximum element from the region of the feature map covered by the filter. Average Pooling computes the average of the elements present in the region of feature map covered by the filter. Thus, while max pooling gives the most prominent feature in a particular patch of the feature map, average pooling gives the average of features present in a patch. 

16. Fully Connected Layer Fully Connected Layer is simply, feed forward neural networks Fully Connected Layers form the last few layers in the network. The  input  to the fully connected layer is the output from the  final  Pooling or Convolutional Layer, which is  flattened  and then fed into the fully connected layer.

17.                Flatten....??? The output from the final (and any) Pooling and Convolutional Layer is a 3-dimensional matrix, to flatten that is to unroll all its values into a vector.

18. Second: Models

19. Covid-19 Image Dataset    3 Way Classification - COVID-19, Viral Pneumonia, Normal

20. About Dataset                     Data was installed from Kaggle.com @ DatasetDownload   Context : Helping Deep Learning and AI Enthusiasts like me to contribute to improving COVID-19 detection using just Chest X-rays. Content : It is a simple directory structure branched into test and train and further branched into the respective 3 classes which contains the images. Acknowledgements : The University of Montreal for releasing the images. Inspiration : Help the medical and researcher community by sharing my work and encourage them to contribute extensively.

21. About Dataset Data was split into test set of total 59 images : Covid : 19 images Normal : 20 images Viral Pneumonia :  20 images Train set of total 251 images : Covid : 111 images Normal : 70 images Viral Pneumonia : 70 images

22. Used Model                              Covid19_Pneumonia and Normal CNN classification python model of accuracy 92% . A python model built on google.colab , where data was installed from kaggle.com and it contained images of corona virus, pneumonia (lung disease) and normal images , the task was to classify these images and  used CNN, the accuracy was 92% . While running this model on google Colab, you have to be signed in to kaggle.com and make a token and upload it, as I am connecting Kaggle with Colab so as not to download the data on my PC. Also at the end saving the model.h5 on google drive , so there will be a connection request to your google drive account to save the model. At the end made a python code to use my .h5 file, i.e. deep learning model to test new data randomly selected from the internet.  Simple Model Deployment.

23. CNN

24.                         Accuracy &  Confusion Matrix

25. Brain MRI Images for Brain Tumor Detection 2 Way Classification – Brain Tumor, No Brain Tumor

26.                About Dataset Data was installed from Kaggle.com @ DatasetDownload Context: Helping Deep Learning and AI Enthusiasts like me to contribute to improving Brain Tumor detection using just MRI. Content: It is a simple directory structure branched into test and train and further branched into the respective 2 classes which contains the images. Inspiration: Help the medical and researcher community by sharing my work and encourage them to contribute extensively.

27. What is a brain MRI?              A brain MRI ( magnetic resonance imaging ) scan, also called a head MRI, is a painless procedure that produces very clear images of the structures inside of your head — mainly, your  brain . MRI uses a large magnet, radio waves and a computer to produce these detailed images. It doesn’t use radiation. Currently,   MRI   is the most sensitive imaging test of your head (particularly, your brain), as compared to other imaging techniques, such as  CT (computed tomography) scans   or  X-rays .

28. About Dataset Data was split into test set of total 210 images : Tumor : 130 images No Tumor  : 80 images Train set of total 43 images : Tumor : 25 images No Tumor  : 18 images

29. Brain Tumor detection model built using CNN and deployed with accuracy 93.02% . A python model built on google.colab, where data was installed from kaggle.com and it contained images of normal brain  and brain with tumor , the task was to classify these images and used CNN, the accuracy was 93.02% . While running this model on google.colab, you have to be signed in to kaggle.com and make a token and upload it, as connecting Kaggle with colab so as not to download the data on my PC. Also at the end  saving the model.h5 on google drive , so there will be a connection request to your google drive account to save the model. At the end made a python code to use .h5 file, i.e. deep learning model to test new data randomly selected from the internet.  Simple Model Deployment.

30. CNN

31. CNN

32. Accuracy

33. Breast Cancer Model

34. Steps of build model 1- Data set (Kaggel-500 normal- 500 sick, ) 2- Preprocessing (Normalization) 3- Modeling:   -Four convolution layers (filter numbers between 64,512 )   -Four pooling layers (max)   -Batch Normalization   -Activation function RELU and Sigmoid in output layer   -validation: accuracy 97.2%

35. Batch Normalization Batch Norm is a normalization technique done between the layers of a Neural Network instead of in the raw data. It is done along mini-batches instead of the full data set. It serves to speed up training and use higher learning rates, making learning easier.

36. Application Description Class Diagram Sequence Diagram

37. Sample of Doctork

38. Sample of Doctork

39. Sample of Doctork

40. Sample of Doctork

41. Let’s Run Visual Studio 2022

42. Brain tumor Classification to 4 classes We used a pre-trained model called Efficient net

43. Efficientnet: Google released a paper in 2019 that dealt with a new family of CNNs i.e EfficientNet. These CNNs not only provide better accuracy but also improve the efficiency of the models by reducing the parameters and FLOPS (Floating Point Operations Per Second) manifold in comparison to the state of art models such as GPipe.

44. To this end, the authors use Neural Architecture Search to build an efficient network architecture, EfficientNet-BO. It achieves 77.3% accuracy on ImageNet with only 5.3M parameters and 0.39B FLOPS. (Resnet-50 provides 76% accuracy with 26M parameters and 4.1B FLOPS).

45. Adversarial Attack Before solve the Adversarial Attack:

46. After apply Fast Gradient Sign Method:

47. Accuracy of the model

48. Testing

49. Testing

50. Testing

51. Skin Cancer Our dataset has two classes (benign & Malignant) Our model achieves 84.5%

52. Architecture of the model:

53. Testing

54. Closure Do you think AI will ever replace humans ?

55. Types of AI The weak AI view holds that intelligent machines cannot really reason and solve problems. These machines only look intelligent, but do not have real intelligence or self-awareness.