https://github.com/axelseancp/cat_breed_classification

A machine learning project for image classification from dicoding machine learning course
https://github.com/axelseancp/cat_breed_classification

computer-vision image-classification machine-learning python tensorflow

Last synced: about 2 months ago
JSON representation

A machine learning project for image classification from dicoding machine learning course

• Host: GitHub
• URL: https://github.com/axelseancp/cat_breed_classification
• Owner: AxelSeanCP
• Created: 2023-12-25T08:10:41.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2024-04-21T13:36:21.000Z (about 2 years ago)
• Last Synced: 2025-02-13T11:54:52.033Z (over 1 year ago)
• Topics: computer-vision, image-classification, machine-learning, python, tensorflow
• Language: Jupyter Notebook
• Homepage:
• Size: 35.1 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Cat_Breed_Classification

A machine learning image classification project to detect cat breeds from dicoding machine learning course

![image of the model prediction result](predict_result.png)

## Dataset

* source : [oxford-iit-cats-extended-10k](https://www.kaggle.com/datasets/doctrinek/oxford-iiit-cats-extended-10k)

* about : this dataset contains 12 differents cat breeds

* samples : 10k images in total from 12 subfolders (class), all the images have different image resolution

## Files

- image_predict → folder of the images used for prediction

- old_code → folder containing old files 

- predict_result → result image file of the prediction

- cat_breed_classification_2.ipynb → notebook file for the project 

- cat_breed_classification_2.py → python file for the project

- vegs.tflite → the saved model ready to be deployed to Android

## Criteria

- free dataset with **1000 images minimum**

- dataset **has never been used** in previous machine learning class submissions

- dataset is split into **80% train set** and **20% test set**

- model should use **model sequential**

- model should use **Conv2D Maxpooling layer**

- accuracy on training and validation set is **80% minimum**

- implement **callback**

- create **plot** for model accuracy and loss

- create code to save model in **TFLite** format

## Optional criteria (5*)

- dataset has minimum **10000 images** and **3 class**

- image resolution in dataset is not constant (varies)

- accuracy on training and validation set is **92% minimum**