A Probabilistic Visual Question Answering Model Based VQA
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Visual data is present everywhere and natural language is a way of communication understandable to humans. Visual Question Answering (VQA) is a system which takes image as an input and a question about the image and generates a natural language answer using complex reasoning. Thus, a VQA needs detailed understanding of the image and complex reason to predict the answer. Given its multimodal structure and possible real-world implementations, VQA is a challenge of critical importance for artificial intelligence. The architectures and hyperparameters used in deep neural networks for VQA have a big impact on their results. This project introduces a pretrained model (VGGNet) to extract image features and Word2Vec to embed the words and LSTM to get word features from the question and after combining the results will predict the answer having highest probability.
Abstract
Visual data is present everywhere and natural language is a way of communication understandable to humans. Visual Question Answering (VQA) is a system which takes image as an input and a question about the image and generates a natural language answer using complex reasoning. Thus, a VQA needs detailed understanding of the image and complex reason to predict the answer. Given its multimodal structure and possible real-world implementations, VQA is a challenge of critical importance for artificial intelligence. The architectures and hyperparameters used in deep neural networks for VQA have a big impact on their results. This project introduces a pretrained model (VGGNet) to extract image features and Word2Vec to embed the words and LSTM to get word features from the question and after combining the results will predict the answer having highest probability.
참고문헌 (17)
- "Introduction to Visual Question Answering: Datasets, Approaches and Evaluation" (2018)[URL]
- Geman Donald, "Visual Turing test for computer vision systems", Proceedings of the National Academy of Sciences of the United States of America (2015)
- Malinowski Mateusz, "A Multi-World Approach to Question Answering about Real-World Scenes based on Uncertain Input", Advances in neural information processing systems (2014)
- Ilievski Ilija, "A Focused Dynamic Attention Model for Visual Question Answering", ~ (2016)
- "3Beginner-Friendly Techniques to Extract Features from Image Data using Python" (2019)[URL]
- Ren Mengye, "Exploring Models and Data for Image Question Answering", ~ (2015)
- Aishwarya Agrawal, "VQA:Visual Question Answering", ~ (2016)
- Yang Zichao, "Stacked Attention Networks for Image Question Answering", 21~29 (2016)
- Qi Wu, "Ask Me Anything: Free-form Visual Question Answering Based on Knowledge from External Sources", ~ (2016)
- Kevin J. Shih, "Where To Look: Focus Regions for Visual Question Answering", ~ (2016)
- Ronghang Hu, "Learning to Reason: End-to-End Module Networks for Visual Question Answering", ~ (2017)
- Yu Jiang, "Pythia v0. 1: the winning entry to the vqa challenge", ~ (2018)
- Akira Fukui, "Multimodal compact bilinear pooling for visual question answering and visual grounding", EMNLP (2016)
- Amanpreet Singh, "Towards VQA Models that can Read", ~ (2019)
- K. Simonyan, "Very deep convolutional networks for large-scale image recognition", ICLR (2015)
- O. Russakovsky, "Imagenet large scale visual recognition challenge", IJCV (2015)
- 이인철, "Exploring Non-finite Verbs of the English Language", 아시아태평양융합연구교류논문지, 7(2), pp.33~43 (2021), 사단법인 한국융합기술연구학회[DOI]