Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/ratikaewkam/pneumoniacnn
Diagnosis of pneumonia using Convolutional Neural Network
https://github.com/ratikaewkam/pneumoniacnn
Last synced: 1 day ago
JSON representation
Diagnosis of pneumonia using Convolutional Neural Network
- Host: GitHub
- URL: https://github.com/ratikaewkam/pneumoniacnn
- Owner: ratikaewkam
- License: mit
- Created: 2024-06-22T10:30:26.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2024-06-26T03:38:50.000Z (5 months ago)
- Last Synced: 2024-06-26T04:42:58.856Z (5 months ago)
- Size: 2.93 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# PneumoniaCNN
Pneumonia Detection Using Convolutional Neural Network
### Idea
I want to build a neural network model that can predict pneumonia. I started by researching neural network architectures, such as AlexNet and VGG. These models use ReLU as their activation function. I wanted to take a different approach and achieve different results, so I began exploring other activation functions like ELU and GELU. Over the course of three months, I designed a new architecture multiple times. Finally, the results were promising.
---
### PXRVision Architecture
- Input (224x224 Grayscale)
- Conv7-s2-64 (ELU) + Conv5-s1-64 (GELU)
- Maxpool3-s2
- Conv3-s1-128 (ELU) + Conv3-s1-128 (GELU)
- Maxpool3-s2
- Conv3-s1-256 (ELU) + Conv3-s1-256 (GELU)
- Maxpool3-s2
- Conv3-s1-512 (ELU) + Conv3-s1-512 (GELU)
- Maxpool3-s2
- Conv3-s1-512 (ELU) + Conv3-s1-512 (GELU)
- Maxpool3-s2
- FC-4096 (ELU) + Dropout 0.5
- FC-4096 (GELU) + Dropout 0.5
- FC-1 (Sigmoid)
### Result
Highest accuracy : 94.87% (Test 5 times)
### References
[Dataset](https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia) |
[Code](https://www.kaggle.com/code/madz2000/pneumonia-detection-using-cnn-92-6-accuracy) |
[VGG](https://arxiv.org/pdf/1409.1556) |
[ELU](https://arxiv.org/pdf/1511.07289) |
[GELU](https://arxiv.org/pdf/1606.08415)
---
Developed by [Rati Kaewkam](https://github.com/ratikaewkam)