# Visual Tree Convolutional Neural Network in Image Classification

## TL;DR

This paper introduces VT-CNN, a novel neural network that uses a confusion-based visual tree to focus on difficult categories, improving classification accuracy on benchmark datasets.

## Contribution

The paper proposes a new Visual Tree Convolutional Neural Network (VT-CNN) that leverages a confusion visual tree to enhance CNN training on challenging categories.

## Key findings

- VT-CNN improves accuracy by up to 1.36% on CIFAR datasets.
- The confusion visual tree effectively identifies difficult categories.
- Experiments demonstrate consistent performance gains over baseline CNNs.

## Abstract

In image classification, Convolutional Neural Network(CNN) models have achieved high performance with the rapid development in deep learning. However, some categories in the image datasets are more difficult to distinguished than others. Improving the classification accuracy on these confused categories is benefit to the overall performance. In this paper, we build a Confusion Visual Tree(CVT) based on the confused semantic level information to identify the confused categories. With the information provided by the CVT, we can lead the CNN training procedure to pay more attention on these confused categories. Therefore, we propose Visual Tree Convolutional Neural Networks(VT-CNN) based on the original deep CNN embedded with our CVT. We evaluate our VT-CNN model on the benchmark datasets CIFAR-10 and CIFAR-100. In our experiments, we build up 3 different VT-CNN models and they obtain improvement over their based CNN models by 1.36%, 0.89% and 0.64%, respectively.