Back to projects

Image Categorisation

Created an image classification pipeline for a 15-class scene dataset using Tiny Images, HOG, SURF Bag of Visual Words, ResNet18 transfer learning, LBP, Random Forest, KNN, SVM, CNN fine-tuning, and confusion matrix analysis.,

SUN images
3000
scene categories
15
methods compared
5

Compared classical image features with deep learning.

The project uses a 3000-image subset of the SUN database across 15 scene categories. Each method extracts visual information differently, then a classifier predicts the scene category from those features.

Accuracy comparison across the best version of each method

The table focuses on what matters for a portfolio reader: the feature method, classifier, accuracy, and what the result shows.

Feature MethodBest ClassifierAccuracyWhat It Shows
Tiny ImageKNN0.322Simple baseline, but compressed images lose important scene detail.
HOGSVM0.503Edge and shape descriptors improved recognition over raw tiny pixels.
SURF BoWSVM0.394Local visual words captured patterns, but were weaker than stronger feature sets.
ResNet18Best overallCNN0.928Best result, using pretrained deep features for high-level scene understanding.
LBPRandom Forest0.551Strongest classical comparison, especially useful for texture-heavy scenes.

Better features produced better classification.

Model training progress

The training output provides supporting evidence from the deep learning experiment, showing how the model learned during the training cycle.

MATLAB training progress output for the Image Categorisation project