Through this project I have created a web application which recognize the fish from Reunion Island's lagoon. The way the application works is simple:
- The user uploads a photo of a fish
- This photo is sent to the Deep Learning algorithm which makes a prediction
- This prediction is returned to the user
Here are the different steps put in place in order to realize this project:
- Collected the data | get_the_data.ipynb
- Trained the model | modeling.ipynb
- Built and deploy a web app with streamlit | app.py
- Create my own dataset by getting pictures from Bing search API
- Use state of the art deep learning models : Resnet50 and apply transfer learning to allow the model to adapt to our problem
- Understand the errors and improve its performance
- Deploy a machine learning application with Streamlit
At first to develop my skills in artificial intelligence and then to spend more time with my grandfather, fan of snorkeling, he spends a lot of time observing the fish. He made me discover his world (🐠) I made him discover mine (🤖)!
After explaining him the project we collected photos in the lagoon together to test the performance of the model in production.
Pictures from our expedition
