Deep Learning Projects

Neural networks, CNNs, and advanced architectures engineered for high-dimensional data processing.

FruitGuard System
CV • YOLOv8 • PyQt5

FruitGuard System

Real-time fruit quality grading system. Multi-stage pipeline: detection (YOLOv8), classification (EfficientNet-B4), and size estimation — achieving sub-100ms inference.

PyTorch OpenCV MLflow Front-end: PyQt5 Back-end: Python
Repo
Generative • VAE

DigitGen VAE

Variational Autoencoder (VAE) implemented for synthetic digit generation. Explores the latent space representation of the MNIST dataset.

PyTorch Latent Space Deep Learning Front-end: Streamlit Back-end: PyTorch
Repo
DigitGen VAE
CV Multitask
HuggingFace • Multi-task

CV Multitask Hub

A unified pipeline for detection, classification, and segmentation using state-of-the-art models from the Hugging Face ecosystem.

Transformers Vision API Front-end: Gradio Back-end: Python
Repo
Hand Tracking • Interaction

AI Gesture Launcher

Innovative system using hand tracking to launch applications and control desktop functions via real-time computer vision gestures.

MediaPipe OpenCV Python Front-end: OpenCV GUI Back-end: Python
Repo
Gesture Launcher
Object Tracking
Deep SORT • Tracking

Deep SORT Tracking

Advanced multi-object tracking system combining detection with deep association metrics for robust object persistence across frames.

YOLOv8 Deep SORT PyTorch Front-end: OpenCV Back-end: Python
Repo
CV • ResNet50

Grape Disease Detection

Automated classification of grape leaves into healthy or diseased categories using a fine-tuned ResNet50 architecture.

Transfer Learning PyTorch Agri-AI Front-end: Streamlit Back-end: Python
Repo
Grape Disease Detection
Real-time Face Detection
CV • OpenCV

Real-time Face & Eye Detection

High-speed computer vision system using Haar Cascade Classifiers for real-time face and eye localization via webcam.

OpenCV Python Real-time Front-end: OpenCV GUI Back-end: Python
Repo