Visual attention-based comparative study on disaster detection from social media images
- PDF / 2,842,879 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 52 Downloads / 181 Views
ORIGINAL ARTICLE
Visual attention-based comparative study on disaster detection from social media images Arif1 · M. Ashraful Amin1 · Amin Ahsan Ali1 · A. K. M. Mahhubur Rahman1 Received: 21 March 2020 / Accepted: 13 July 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract The availability of images of events almost in real-time on social media has a prospect in many application developments. A humanitarian technology for disaster type and level assessment can be developed using the images and video available on social media. In this paper, we investigate the potential use of various available deep learning techniques to develop such an application. For our research, based on the use of publicly available image data, we have started collecting disaster images from various sources from South Asia. We created the South Asia Disaster (SAD) image dataset containing 493 images from various online news portals. Using the Keras as our framework to run our models: Visual Geometry Group (VGG-16 and VGG-19), Inception-V3 and Inception-ResNet-V2 (ResNet: Residual Network). However, to boost up the training speed, we dropped the fully connected layer and added a small, fully connected model. To identify the five different disasters: fire disaster, flood disaster, human disaster, infrastructure disaster, natural disaster; our proposed method with VGG-16 model’s recognition accuracy was 84.51%, which is the highest accuracy on the SAD dataset. After performing the testing, we calculate the VGG-16 classifier’s attention to visualize which part of the disaster images VGG-16 pays attention. Keywords Disaster image · Humanitarian technology · Standard disaster dataset · Convolutional neural network · VGG-16 · VGG-19 · Inception-V3 · Inception-ResNet-V2 · Keras model · Visual attention
1 Introduction Bangladesh is one of the most vulnerable countries that suffer from huge climate change, as well as disasters. Its population density and socioeconomic environments make it highly susceptible to many human-made hazards that include fire, collapse building, infrastructural damage, road accident, etc. Also, there are other disasters happening like infrastructure or non-natural disasters. Day after day, people are suffering from different kinds of disasters (e.g., fire incidents in
B
A. K. M. Mahhubur Rahman [email protected] http://www.agencylab.org Arif [email protected] M. Ashraful Amin [email protected] Amin Ahsan Ali [email protected]
1
Artificial Intelligence and Cybernetics Lab, CSE, Independent University, Bangladesh, Dhaka, Bangladesh
Chawk Bazar, Banani Fire incident). These incidents which happened in March 2019 were featured by some newspapers such as the famous online news portal Dhaka Tribune [20], The Daily Star [5]. It is crucial in times of crisis that how emergency response workers reach all those affected promptly. It would be great to have a system that would raise an alert and determine the degree of damage of any disaster and inform the appropriate authorities based on the aut
Data Loading...
