The Inception Team at VQA-Med 2020: Pretrained VGG with Data Augmentation for Medical VQA and VQG

Aisha Al-Sadi; Hana Al-Theiabat; Mahmoud Al-Ayyoub

The Inception Team at VQA-Med 2020: Pretrained VGG with Data Augmentation for Medical VQA and VQG

Aisha Al-Sadi
, Hana Al-Theiabat
, Mahmoud Al-Ayyoub

Jordan University of Science and Technology

Research output: Contribution to journal › Conference article › peer-review

7 Scopus citations

Abstract

This paper describes the methodology of The Inception team participation at ImageCLEF Medical 2020 tasks: Visual Question Answering (VQA) and Visual Question Generation (VQG). Based on the data type and structure of the dataset, both tasks are treated as image classification tasks and are handled by using the VGG16 pre-trained model along with a data augmentation technique. In both tasks, our best approach achieves the second place with an accuracy of 48% in the VQA task and a BLEU score of 33.9% in the VQG task.

Original language	English
Journal	CEUR Workshop Proceedings
Volume	2696
State	Published - 2020
Externally published	Yes
Event	11th Conference and Labs of the Evaluation Forum, CLEF 2020 - virtual, Online, Greece Duration: 22 Sep 2020 → 25 Sep 2020

Keywords

Augmentation
ImageCLEF 2020· VQA-Med· Visual Question Answering
Medical Image Interpretation
Medical Questions and Answers
Transfer Learning
VGG Network
Visual Question Generation

Cite this

@article{b9601ea685274cdd862866c66e323186,

title = "The Inception Team at VQA-Med 2020: Pretrained VGG with Data Augmentation for Medical VQA and VQG",

abstract = "This paper describes the methodology of The Inception team participation at ImageCLEF Medical 2020 tasks: Visual Question Answering (VQA) and Visual Question Generation (VQG). Based on the data type and structure of the dataset, both tasks are treated as image classification tasks and are handled by using the VGG16 pre-trained model along with a data augmentation technique. In both tasks, our best approach achieves the second place with an accuracy of 48\% in the VQA task and a BLEU score of 33.9\% in the VQG task.",

keywords = "Augmentation, ImageCLEF 2020· VQA-Med· Visual Question Answering, Medical Image Interpretation, Medical Questions and Answers, Transfer Learning, VGG Network, Visual Question Generation",

author = "Aisha Al-Sadi and Hana Al-Theiabat and Mahmoud Al-Ayyoub",

note = "Publisher Copyright: Copyright {\textcopyright} 2020 for this paper by its authors.; 11th Conference and Labs of the Evaluation Forum, CLEF 2020 ; Conference date: 22-09-2020 Through 25-09-2020",

year = "2020",

language = "English",

volume = "2696",

journal = "CEUR Workshop Proceedings",

issn = "1613-0073",

publisher = "CEUR-WS",

}

TY - JOUR

T1 - The Inception Team at VQA-Med 2020

T2 - 11th Conference and Labs of the Evaluation Forum, CLEF 2020

AU - Al-Sadi, Aisha

AU - Al-Theiabat, Hana

AU - Al-Ayyoub, Mahmoud

PY - 2020

Y1 - 2020

N2 - This paper describes the methodology of The Inception team participation at ImageCLEF Medical 2020 tasks: Visual Question Answering (VQA) and Visual Question Generation (VQG). Based on the data type and structure of the dataset, both tasks are treated as image classification tasks and are handled by using the VGG16 pre-trained model along with a data augmentation technique. In both tasks, our best approach achieves the second place with an accuracy of 48% in the VQA task and a BLEU score of 33.9% in the VQG task.

AB - This paper describes the methodology of The Inception team participation at ImageCLEF Medical 2020 tasks: Visual Question Answering (VQA) and Visual Question Generation (VQG). Based on the data type and structure of the dataset, both tasks are treated as image classification tasks and are handled by using the VGG16 pre-trained model along with a data augmentation technique. In both tasks, our best approach achieves the second place with an accuracy of 48% in the VQA task and a BLEU score of 33.9% in the VQG task.

KW - Augmentation

KW - ImageCLEF 2020· VQA-Med· Visual Question Answering

KW - Medical Image Interpretation

KW - Medical Questions and Answers

KW - Transfer Learning

KW - VGG Network

KW - Visual Question Generation

UR - https://www.scopus.com/pages/publications/85121837181

M3 - Conference article

AN - SCOPUS:85121837181

SN - 1613-0073

VL - 2696

JO - CEUR Workshop Proceedings

JF - CEUR Workshop Proceedings

Y2 - 22 September 2020 through 25 September 2020

ER -

The Inception Team at VQA-Med 2020: Pretrained VGG with Data Augmentation for Medical VQA and VQG

Abstract

Keywords

Other files and links

Fingerprint

Cite this