Dual-Branch Dynamic Graph Convolutional Network for Robust Multi-Label Image Classification

Abstract

For the intricate task of multi-label image classification, this paper introduces an innovative approach: an attention-guided dual-branch dynamic graph convolutional network. This methodology is designed to address the difficulties faced by current models when handling multiple labels within images. By integrating multi-scale features, it enhances the retention of original category information and boosts the robustness of feature learning. Utilizing a semantic attention module, the study dynamically reweights feature categories in the training dataset, enhancing the network's capability to identify smaller objects and generate context-sensitive category representations. The effectiveness of the proposed model was evaluated using the MS-COCO2014 imagery dataset, demonstrating superior performance in critical metrics such as classification precision (CP), recall (CR), and F1 score (CF1), outperforming other state-of-the-art models. Furthermore, a cascaded classification structure was implemented to leverage the prior information from static images to inform the processing of dynamic ones, and to utilize original image category data to augment label correlations, thereby enhancing overall classification accuracy.

Authors and Affiliations

Bingxing Wang, Hongye Zheng, Yingbin Liang, Guanming Huang, Junliang Du

Keywords

Related Articles

A Study on Channel Access Algorithms for the Cellular Systems

Recently, there has been a lot of focus on channel distribution and handoff techniques for cell structures in order to provide long-term organizations that provide QoS to convenient customers. We have a well-organized di...

Twitter Data Classification by Applying and Comparing Multiple Machine Learning Techniques

Having an average of five hundred million tweets sent out per day, twitter has become one of the largest platforms of data analysis for the researchers. Previously, various researches have been conducted on twitter data...

A Review of Laboratory Studies on Use of Low Density Polyethene (LDPE) and Non-Silica as Bitumen Modifiers

Pavements experience numerous distresses including rutting, fatigue cracking, ravelling, etc. as a result of rising axle loads, rising traffic volumes, and poor maintenance. These annoyances lengthen travel times and rai...

Dynamic Modelling and Control Analysis of a Fuel Cell Connected to Electric Vehicle

A fuel cell electric vehicle (FCEV) is an electric vehicle that uses a fuel cell to power an electric motor. FCEV aims to provide customers with the benefits of battery electric vehicles such as low-to-zero emissions, hi...

A Literature Review on Big Data Analytics

Huge volumes of data have been available to policymakers in the digital world. Big data is a term to collections that are not always huge, but also varied and fast changing, rendering standard tools and procedures inadeq...

Download PDF file
  • EP ID EP749876
  • DOI 10.55524/ijircst.2024.12.5.13
  • Views 76
  • Downloads 0

How To Cite

Bingxing Wang, Hongye Zheng, Yingbin Liang, Guanming Huang, Junliang Du (2024). Dual-Branch Dynamic Graph Convolutional Network for Robust Multi-Label Image Classification. International Journal of Innovative Research in Computer Science and Technology, 12(5), -. https://www.europub.co.uk/articles/-A-749876