Android studio Environment for development of the application
Training Environment for the Image classification Model
Icon for the Android application
Testing the Performance of the image classification model in Azure custom Vision
Menu page of the Android application
Second page of the Android application
Screenshot of the result of my first training on the images
Result of the second training on Images
Result of third training
Result of fourth training with a model performance of 99.8%
Basically, when faced with the challenge of coming up with a final year project as requirement to fulfillment of my BSc pursuit in Systems Engineering, I had no idea what kind of project i can work on, so i made some research and did some findings where I discovered that in some rural and under-developed areas in the world today, during malaria diagnosis by a skilled and experienced lab scientist, manual counting of presence of malaria parasite in a cell is still done just to verify if a patient is infected with malaria or not.
What it does
It scans images of cells to detect if the cell is infected with malaria or not, either when connected to a standard microscope in the lab during diagnosis or when the image of a cell is scanned from almost any surface.
How I built it
- I got the malaria cell images data-set from Kaggle
- I examined the images i got and separated them into different categories for training, testing and for validation 3.I setup my Microsoft Azure environment by registering and signing up for azure services.
- I logged on to my Microsoft Azure portal, and created a resource for the project.
- I then logged on to link where I loaded my images and trained my classification model or image recognition system.
- I exported my trained model for deployment into Android as an android application by exporting the model for further development in Android studio using Java programming language to build the application.
- I developed the android application and generated the signed apk for it, so as to test it and make it available for use.
Challenges I ran into
- Signing up for Azure account and getting it verified
- Exporting the trained model from Azure Custom Vision to Android studio
- Understanding the original sample code after importing models from Azure custom vision.
- Developing and customizing my Android application
Accomplishments that I'm proud of
- I was able to build my own image recognition system or AI using Microsoft Custom Vision
- Being able to export the trained model
- Being able to import this model into Android studio for further development
- Finally able to fully develop the application as a means of deploying the trained Model for malaria cells detection.
What I learned
- Using Microsoft Azure Custom Vision Services
- Training Image recognition models
- Being able to understand Artificial intelligence concepts and Data science concepts too.
- Understanding the process of Malaria Diagnosis
- Having a full understanding of how malaria threatens man kind in different parts of the world
What's next for PLASMO-D
- I intend making the android application more interactive and user-friendly
- I plan on making it available to everyone who wants access to it. 3, If possible i intend making it an application that can be used for various sickness diagnosis