Fish scales constitute a valuable source of information about individual life histories, but correctly extracting this information requires a highly skilled expert. Here, we train a deep convolutional neural network architecture EfficientNet B4 on a set of about 9000 salmon scale images, and show that it attains good performance on predicting a …

StageMamba is a deep learning framework for multi-class eye disease classification from retinal fundus images. It combines EfficientNet-B4, Multi-Level Feature Fusion (MLFF), and Mamba-based Vision State Space blocks to capture both local retinal details and global disease patterns, achieving robust and efficient diagnosis across 10 eye disease

Repository to archive and document work done for satisfying Capstone Project requirements at Praxis Business School.

AI-powered Deepfake Detection System with Explainable AI (Grad-CAM), built using FastAPI, Next.js, and EfficientNet-B4. Supports image & video analysis with high accuracy (96%+) and real-time inference.