I’ve done self-learning on AI and machine learning for nearly two years and I found out that doing projects is very important for self-learning, because it helps us to solve the real problem and get our hands dirty in the technology that we are learning.
I want to share the process of the project with everyone who just start in AI and ML so you can have a sense of an end to end process of any AI project. I will show you every step in detail from data collection, data preparation, data processing, feature engineering, model training, model deployment, and how to connect any client to the model that I’ve built.
This project, I will divide into 3parts:
- Project overview and Architecture: I will give an overview of my application, how I built it, and also the technology that will be used in the project
- Model Training: I will include data collection, data preprocessing, model training using Keras, and transfer learning
- Model deployment: You can deploy your model by using flask, fast API, or Django, but in my case I want my model to work offline so I will show you how you can convert the model into TensorFlow lite to use in a mobile app.
1. Project Overview
The institute where I study is located on the mountain and in the rainy season there are a lot of mushrooms, and sometimes we want to collect the mushrooms from the forest but we are not sure which mushroom is dangerous or poisonous. Because of this problem I came up with the idea which is building a mobile application that can identify the mushroom and also tell the user if it is edible or not.
2. Project Architecture
In this project, I will download the mushroom images from Kaggle, which has around 16 types of mushroom, then use those data to train a TensorFlow model using transfer learning, After training, I do quantization on the model to reduce the memory of the model that make it lightweight for use in mobile application
dataset: https://www.kaggle.com/datasets/chipprogrammer/mashroom-image-classification
3. Conclusion
In this article, you already know the overview of the project and also the Architecture of the application, with my plan I will explain to you in detail how I build the model, measure the performance of the model also the deployment and mobile application development in the next article.
