Learning visual concepts in images using temporal convolutional networks

Qian, Chen, Chaturvedi, Iti, Poria, Soujanya, Cambria, Erik, and Malandri, Lorenzo (2019) Learning visual concepts in images using temporal convolutional networks. In: Proceedings of the 2018 IEEE Symposium Series on Computational Intelligence, SSCI 2018. 8628703. pp. 1280-1284. From: SSCI 2018: IEEE Symposium Series on Computational Intelligence, 18-21 November 2018, Bengalaru, India.

[img] PDF (Published Version) - Published Version
Restricted to Repository staff only

View at Publisher Website: https://doi.org/10.1109/SSCI.2018.862870...
 
3
1


Abstract

With the advancement of technology and the rapid rise of social media, people now tend to share their experiences and feelings using images. It is useful to effectively mine opinions and identify sentiments from the images available on various platforms for tasks like social media marketing and user profiling. Visual sentiment concepts extracted from tags of online images are known as adjective noun pairs and the sentiment conveyed by them is usually predicted using opinion mining methods. However, the pre-training work is time consuming and requires huge space. Instead of collecting tags for the images, in this paper we propose to automatically predict the corresponding sentiments using deep convolutional networks. Next, to model the sequence of different sentiments in a single image we consider a recurrent neural network. Such a model is able to remember the context of different sentiments in a single image. We applied our method on Flickr dataset and our approach outperformed baselines in the range of 3-20%.

Item ID: 63343
Item Type: Conference Item (Research - E1)
ISBN: 978-1-5386-9276-9
Keywords: Image Sentiment Analysis, Temporal Convolutional Networks
Copyright Information: © 2018 IEEE.
Date Deposited: 06 Jul 2020 22:34
FoR Codes: 46 INFORMATION AND COMPUTING SCIENCES > 4603 Computer vision and multimedia computation > 460306 Image processing @ 100%
SEO Codes: 97 EXPANDING KNOWLEDGE > 970108 Expanding Knowledge in the Information and Computing Sciences @ 100%
Downloads: Total: 1
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page