Imbalanced Data Problem in INTERCO Detection

doi:10.1007/978-3-032-09321-9_33

Return

Imbalanced Data Problem in INTERCO Detection

2026-01-01

PRE

Juliusz Łosiński *

Ireneusz Czarnowski

DOI：10.1007/978-3-032-09321-9_33

Original

Original request for help

Save

Abstract

En 中文

The problem of imbalanced data appears when the number of instances in one class is significantly greater than the number of instances in other classes. This situation leads to challenges in developing robust machine learning models. To solve this problem, various algorithms and strategies, such as undersampling and oversampling, have been proposed to achieve a balanced dataset by either reducing or increasing the instances of a particular class in a dataset. In the field of computer vision, the problem of imbalanced data is particularly frequent, as some object classes often have fewer labelled images compared to others, posing challenges for object detection models like YOLO. This paper addresses the challenge of class imbalance in the context of detecting signals from the International Code of Signals (INTERCO). This paper involves the use of state-of-the-art object detection models, YOLOv8, YOLOv9, YOLOv10, and YOLOv11, combined with data balancing techniques to enhance performance. The aim of the balancer methods was to increase the representation of minority classes and eliminate the imbalanced classes problem, consequently enhancing the detection accuracy. In order to balance the dataset, methods such as data augmentation, autoencoders, deep convolutional generative adversarial networks (DCGANs), the synthetic minority oversampling technique (SMOTE), and adaptive synthetic sampling (ADASYN) have been used. The computational experiment results of these methods are then presented and discussed.

Keywords:

Imbalanced dataset

Synthetic data

Computer vision

YOLO