I built an AI image classifier application to recognize different species of flowers using Pytorch, then convert it into a command line application as a requirement for Udacity's AI Programming with Python Nanodegree program.
- This AI application was trained on a dataset of 102 flower categories gotten from ImageNet. Using Transfer Learning i built a model which uses a deep learning model trained on hundreds of thousands of images as part of the overall application architecture.
- It identifies name of Flowers, going forward AI algorithms will be incorporated into more and more everday applications. This model can be integrated into a phone app that tells you the name of the flower your camera is looking at.
- Pytorch
- Jupyter Notebook
- Google Colab
- Command Line application
- Ready trained model
- It is required that Python is already installed.
- Also note training data is not included in this repo
- Required dependencies are located in a file requirements.txt
pip install -r requirements.txt
- Clone this repo or download as zip file
- Open a commandline prompt, navigate to the folder directory
cd C:/Users/emmas/Desktop/emmanuel_udacity - Train the model, replace data_directory with flower dataset(Prints out training loss, validation loss and validation accuracy as it trains)
python train.py data_directory - Some options include:
- Set directory to save checkpoints:
python train.py data_dir --save_dir save_directory - Set hyperparameters:
python train.py data_dir --learning_rate 0.01 – hidden_units 512 – epochs 20 - Use GPU for training:
python train.py data_dir --gpu - Choose architecture:
python train.py data_dir --arch
- Set directory to save checkpoints:
- Predict a flower name from a single image path, this returns the flower name and class probability:
python predict.py /path/to/image checkpoint - Options include:
- Return top K most likely classes:
python predict.py input checkpoint --top_k 3 - Use a mapping of categories to real names:
python predict.py input checkpoint --category_names cat_to_name.json - Use GPU for inference:
python predict.py input checkpoint --gpu
- Return top K most likely classes:
Project is: complete
Room for improvement:
- This model can be further integrated into a Phone App, that uses the camera to identify and display name of flowers.
- This project was based on a requirement for Udacity's AI Programming with Python Nanodegree program.
- Many thanks to Udacity and AWS for the opportunity.
Created by [@Emmanuel-Samuel] - feel free to contact me!
This project is available under the Udacity License.