May 23, 2022 · We propose VQA-GNN, a new VQA model that performs bidirectional fusion between unstructured and structured multimodal knowledge to obtain unified knowledge ...

The visual question answering (VQA) task aims to pro- vide answers to questions about a visual scene. It is cru- cial in many real-world tasks including ...

This work proposes a novel visual question answering method, VQA-GNN, which unifies the image-level information and conceptual knowledge to perform joint ...

May 23, 2022 · In this work, we propose a novel visual question answering method, VQA-GNN, which unifies the image-level information and conceptual knowledge ...

These two reasoning paths demonstrate that VQA-GNN is an inoperable method that can give a reasonable explanation to each choice with our well structured ...

Missing: Semantic Answering.

May 23, 2022 · In this work, we propose a novel visual question answering method, VQA-GNN, which unifies the image-level information and conceptual knowledge ...

The VQA-GNN model [23] extended this approach to the VQA domain by incorporating image concepts and multi-modal data, although it only considers knowledge ...

Jul 22, 2024 · Connected Papers is a visual tool to help researchers and applied scientists find academic papers relevant to their field of work.

By overlaying this process multiple times, the model performs iterative inference and predicts the optimal answer by analysing all problem-oriented evidence.