README.md

ECG Heartbeat Classification with 1D-CNN

This project aimed to classify the heart beat types with the 1D-CNN model trained with the MIT-BIH dataset.

Current status

Status	Act
Finished	Data preprocess
Finished	Data analysis
Finished	Base model created
On progress	Model improvement
	Testing on different datasets

Project Directory Hierarchy

Project hierarchy is below with subfolders and files:

 ecg-anomaly-detection 
 	 |-- LICENSE
 	 |-- README.md
	 |-- requirements.txt
	 |-- main.py
	 |-- train.py
	 |-- wave_visualize.py
	 \-- model
	   |-- __init__.py
	   |-- model.py
	   \-- trained_models
	     |-- (trained models, going to be saved here)
	   \-- logs
	     |-- (tensorboard logs, going to be saved here)
	 \-- data
	   |-- __init__.py
	   |-- data_generator.py
	   |-- txt2csv.py
	   |-- convert_all_txt2csv.py
	   \-- raw
	     |-- 100.csv
	     |-- 101.csv
	     |-- (45 more files ...)
	     |-- 234.csv
	   \-- annotations
	     \-- csv
	       |-- 100annotations.csv
	       |-- 101annotations.csv
	       |-- (45 more files ...)
	       |-- 234annotations.csv
	     \-- txt
	       |-- 100annotations.txt
	       |-- 101annotations.txt
	       |-- (45 more files ...)
	       |-- 234annotations.txt

Requirements

• Tested on Ubuntu 18.04 and with Python 3.7.x
• Anaconda 4.8.3 (or Miniconda)
• Python Libraries
  • NumPy
  • Tensorflow 2.x
  • Matplotlib
  • Scikit-Learn
  • Pandas
  • PyQT5 (for Wave Visualizing)
  • Tqdm

Heartbeat Wave Visualizing

Sample monitor is above.

Cloud link of project files

• ✔️ Data files uploaded
• ✔️ Trained models uploaded

Link: https://drive.google.com/open?id=188bqrXPn23Ad6FJxwbBDyfhJpo8rwUdV

Contact

For more further questions

• E-mail: omerf.sarioglu@gmail.com
• LinkedIn: omerfsarioglu