Sitraka R.

This is an Android application developed combining computer vision and geolocation to guide users. It uses the ORB (Oriented FAST and Rotated BRIEF) algorithm with FLANN matching and RANSAC homography to identify campus buildings in real-time via the camera. The application also integrates an interactive map based on OpenStreetMap, allowing exploration of surrounding buildings with their real-time distances, as well as a GPS navigation module to guide the user to their destination. Map data is extracted from the PBF OpenStreetMap file for the Toamasina region using osmium-tool, ensuring accurate campus coverage.

Cerebra AI is an Android mobile application for classifying brain MRI images, developed to automatically detect 4 types of pathologies (glioma, meningioma, pituitary tumor, and absence of tumor). It embeds an optimized MobileNetV3Small model converted to TensorFlow Lite, allowing for fast (≈18 ms) and entirely offline inference. The application offers an ergonomic interface with image preprocessing (sharpening + histogram equalization), analysis saving, PDF report export, and a physician diagnostic module for clear traceability. Intended for experimental use, it illustrates the integration of embedded artificial intelligence into a medical decision support tool.