Cat_Classifier

Overview

An image classifier which can classify 20 different breeds of cats based on their image.Those breeds are

1. Abyssinian cat
2. Aegean cat
3. Balinese cat
4. Bengal cat
5. Birman cat
6. Bombay cat
7. British Longhair cat
8. Burmese cat
9. Burmilla cat
10. Cornish Rex cat
11. Cymric cat
12. Donskoy cat
13. Oregon Rex cat
14. Oriental Bicolor cat
15. Persian cat
16. Pixie-Bob cat
17. Ragamuffin cat
18. Siamese cat
19. Siberian cat
20. Turkish Angora cat

Table of Contents

• Installation
• Dependencies
• Data Collection
• Data Modelling
• HuggingFace Deployment

Installation

To get started with this project, follow these steps:

1. Clone the repository:

   git clone https://github.com/marufc36/Cat_Classifier.git

Dependencies

pip3 install requirements.txt

Data Collection

In my cat image classifier project, I utilized Python code and the DuckDuckGo browser to efficiently download a diverse dataset of cat images. This approach allowed me to gather a comprehensive collection, ensuring the robustness and accuracy of my classifier in recognizing and categorizing different feline breeds.

https://colab.research.google.com/drive/1U5Jh2-r1TapSpMT1N2gu77njrEyEnH2G?usp=sharing

Training Batch

Validation Batch

Data Modelling

Using fastai's vision learner and restnet34 model my model have achieved accuracy over 80%.Confusion matrix is given bellow.

https://colab.research.google.com/drive/1Oe52UDJEzPfvNyTb0TfXdk9YlMGoxHgW?usp=sharing

HuggingFace Deployment

This model deployed into HuggingFace Spaces. Link is given bellow.

https://huggingface.co/spaces/mmchowdhury/CatClassifier