← Back to Projects

CNN Image Classifier for Household Objects

Deep learning model achieving 93.4% accuracy in classifying 4 household object types

Overview

Implemented Convolutional Neural Network with Transfer Learning (ResNet18) for classifying four household objects: cooking pots, cups, knives, and pens. Achieved 93.4% overall accuracy with comprehensive data augmentation, rigorous evaluation, and detailed per-class performance analysis. The model demonstrates particularly strong performance on cooking pots (98.4%) and cups (96.8%).

Problem

Need for accurate automated classification of household objects for inventory management and object recognition applications

Solution

Implemented CNN with Transfer Learning (ResNet18), data augmentation, class-specific optimization, and rigorous evaluation metrics

Results

Overall Test Accuracy: 93.4%

Cooking Pot: 98.4% accuracy

Cup: 96.8% accuracy

Pen: 94.6% accuracy

Knife: 93.1% accuracy

Comprehensive confusion matrix analysis

Robust performance across all object classes

Technologies

PythonPyTorchtorchvisionscikit-learnOpenCVTensorBoard

Related Projects

View Project

AI/ML

Music Generation with RNNs

Deep learning system generating Irish folk music using 2-Layer LSTM trained on 214K tunes

Python
PyTorch
Flask
+4