The objective of this repository is to understand how to integrate and use machine learning models with Flutter.
A Flutter app that detects a type of leaf given a photo of an unknown leaf.
You can download the latest version of APK LINK.
- Clone or download the repository.
- Open it in VSCode or Android Studio
- Go to pubspec.yaml and click on 'Pub get' to download all the dependencies used in project.
- Use and Modify the project as per your choice.
- On launching the application, you will be presented with the instructions column.
- The application then runs the TFLITE model in the background to get the name of unknown leaf.
- It displays the results on the next screen
Suggestions
.
- The
tflite
model has been trained to detect only a subset of the leaves. This includes:- Apple
- Grape
- Peach
- Pepper
- Potato
- Tomato
- Not Valid
- The size of the dataset was around 5,000 photos which is sufficient enough to make the model recognize selected
- The dataset was taken from KAGGLE.
Contributions towards the project are welcome. Specifically:
- The
tflite
model used can be replaced with a more accurate one/one with more leaves. - The responsiveness of the application can be improved.
- The app can be trained on a more extensive dataset to include more types of leaves and their diseases.
- The requirements I set have been made possible. However, the application can still be improved.
- Additions/improvements can be made as specified in the
Contributing
section.