https://github.com/hawmex/aut_ai_project

This repository contains the files of my project for the "Artificial Intelligence" course at AUT (Tehran Polytechnic).
https://github.com/hawmex/aut_ai_project

classification cnn computer-vision machine-learning python

Last synced: 3 months ago
JSON representation

This repository contains the files of my project for the "Artificial Intelligence" course at AUT (Tehran Polytechnic).

• Host: GitHub
• URL: https://github.com/hawmex/aut_ai_project
• Owner: Hawmex
• License: mit
• Created: 2023-01-02T08:30:18.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2024-03-28T10:59:13.000Z (over 1 year ago)
• Last Synced: 2025-02-12T07:22:22.704Z (5 months ago)
• Topics: classification, cnn, computer-vision, machine-learning, python
• Language: Jupyter Notebook
• Homepage:
• Size: 9.53 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Defective Pump Impeller Detection Using a CNN

This repository contains the files of my project for the "Artificial

Intelligence" course at Amirkabir University of Technology (Tehran Polytechnic).

## Project Description

This project deals with a classification problem. Since we are trying to

classify images, we have decided to solve it with a convolutional neural network

due to its abilities in image processing.

## Dataset

The data we used are available on

[Kaggle](https://www.kaggle.com/ravirajsinh45/real-life-industrial-dataset-of-casting-product).

### Examples

#### Defective Pump Impeller

![Defective Pump Impeller](./screenshots/defective_pump_impeller.png)

#### Non-Defective Pump Impeller

![Non-Defective Pump Impeller](./screenshots/non_defective_pump_impeller.png)

## The Convolutional Neural Network

We created a CNN with the following architecture:

- `Conv2D`: 64x64x8

- `MaxPool`: 32x32x8

- `Conv2D` 32x32x8

- `MaxPool`: 16x16x8

- `Flatten`: 2048

- `Dense`: 16

- `Dense`: 16

- `Dense`: 1

Total params: 33,737

![The Convolutional Neural Network](./screenshots/model.png)

## Training and Validation

### Accuracy

We used `binary_accuracy` as the performance metric of the network.

![Accuracy](./screenshots/accuracies.png)

### Loss

We used `binary_crossentropy` as the loss function of the network.

![Loss](./screenshots/losses.png)

## Results

Our CNN model can classify images of pump impeller with an accuracy of **~96%**.