Sricharan Koride

Image Classification Using Transfer Learning

This project builds an image classifier using transfer learning with pre-trained models like ResNet50, EfficientNetB0, and VGG16. It fine-tunes only the final layer while freezing earlier layers to extract meaningful features. Data augmentation (cropping, rotation, flipping) enhances generalization, and the model is trained with ReLU activation, batch normalization, dropout, and ADAM optimizer for at least 50 epochs with early stopping. The goal is to create a robust classifier optimized for small datasets using state-of-the-art deep learning techniques.

• Python
• TensorFlow
• Keras
• PyTorch
• OpenCV
• NumPy
• Pandas
• Matplotlib
• Scikit-learn

CSC Grameen-E Works

Developed during my undergrad, this is the user module of a mobile app that enables users to book a host of services in rural areas, developed pursuant to Digital India's mission of expanding internet connectivity to everyone.

• Android Studio
• Java

Speech Emotion Recognition with Librosa

Speech Emotion Recognition with Librosa was developed as a mini-project during undergraduate studies at CVR College of Engineering. The project focuses on recognizing human emotions from speech using machine learning techniques. It extracts features from audio files, and classifies emotions like happiness, sadness, anger, and fear. The model is trained on the Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) dataset and utilizes a Multi-Layer Perceptron (MLP) classifier for emotion prediction. The system includes a Flask-based web interface, allowing users to upload an audio file and receive a predicted emotion label in real time.

• Python
• Librosa
• Scikit-Learn
• NumPy
• HTML5, CSS, Bootstrap
• Pickle
• Soundfile

A Study of Post-Quantum Cryptopgraphy in TLS Integration

As part of CSCI-520 Security Systems course at USC, I authored this survey paper that explores various efforts to integrate Post-Quantum Cryptopgraphic algorithms into TLS 1.3 to safeguard it from quantum cryptanalytic attacks. This was praised by Professor Clifford Neuman, a director at ISI as "an extremely well written and well organized paper on what is often a very confusing topic. Good job."

View Paper →

Color Identification System for Visually Impaired People

Developed as part of our undergraduate major project, this system helps visually impaired individuals identify colors independently. The app detects colors in real-time from a live camera feed and provides both visual and audio feedback. It features color detection (identifying the color of a selected point) and color search (highlighting all areas with a specific color). Built for accessibility, the project was published in the International Journal of Scientific Engineering and Applied Science (IJSEAS) and has potential applications in assistive technology, object tracking, and automation.

• Python
• NumPy
• Pandas
• Kivy
• OpenCV

Visit IJSEAS →