3D Brain Tumor Detection Using Vision Transformers and Explainable AI

Brain tumor detection using Vision Transformers (ViT) for highly accurate image analysis, combined with Explainable AI (XAI) techniques such as Grad-CAM, LIME, and SHAP to provide transparency and interpretability in predictions.

AU-Net model is employed for precise segmentation of brain tumors, with advanced 3D visualizations, dynamic colored graphs, and GIFs used to enhance representation.

Minimal NeRF Implementation with JAX and Flax

Implementation of Neural Radiance Fields (NeRF) using the JAX and Flax libraries. NeRF is a technique for generating 3D scene representations from 2D images by modeling how light interacts with objects to create photorealistic renders.

Emotion Recognition from Audio Speech

Speech Emotion Recognition (SER) model using LSTM layers, NLP techniques, and deep learning to classify speech recordings into six emotions: fear, angry, disgust, neutral, sad, and happy.

Video Anomaly Detection

Video processing project using DenseNet121 Transfer Learning to detect 14 categories of abnormal situations in video with an accuracy of 91.53%.

Agro Tech Nexus

Developed an all-in-one solution for farmers, including crop-price prediction, weather alerts, cold-storage information, market access transportation, a tool marketplace, and chatbot/phone-based assistance.

CNN Model for Traffic Images

Computer Vision project to identify traffic images and signals using both Rectified Linear Unit (ReLU) and Softmax activation functions. The model achieved an accuracy of 96%.

This was a project worked on at NIT-C.

Phishing Website Detection Using Machine Learning SVM Linear Classifier

Developed a Machine Learning model for detecting phishing websites utilizing a Linear Support Vector Machine. The model achieved a robust accuracy of 94%.

This was a project worked on at NIT-K.

Solar Panel Fault Detection Using CNN & VGG16

Developed a deep learning model using Convolutional Neural Networks (CNN) and VGG16 architecture to detect faults in solar panels. The model is designed for highly accurate fault detection through image analysis, contributing to efficient solar panel maintenance and energy output.