Understanding Domain Adaptation Using CORAL in Computer Vision
Aditya Chakraborty
Affiliation: The Village School
IJSCAR Vol. 2, Issue 2 (2025) · pp. 28–36
Abstract
This paper investigates whether Domain Adaptation techniques can significantly improve the performance of Convolutional Neural Networks (CNNs) in image classification across varying domains and distributions. Our work applies Deep CORAL with EfficientNetV2 for domain adaptation on the Office-31 dataset. We compare its performance to a regular EfficientNetV2 model that doesn’t use domain adaptation measuring improvements with metrics as follows: accuracy precision recall and F1 score. The CORAL-implemented model demonstrated a 4.15% average boost in average precision recall F1 and accuracy across all 3 trials.
Keywords: Artificial Intelligence, Computer Vision, Deep Learning, Domain Adaptation, Neural Networks, Convolutional Neural Networks, Transfer Learning