Image Caption Generation in Telugu

Numerous methods are available to describe images by words and image captioning is a well-examined and researched topic. Further, the subject was much studied to work with English. Our model functions to generate descriptive explanations for general images in Telugu using Deep Learning techniques. T...

Full description

Saved in:
Bibliographic Details
Main Authors: Dwarampudi, Jagan Mohan Reddy, Rampavan, Damerla, Sathwik, Mandavilli Anu Sandeep, Reddy, Kaipu Nivas, Mishra, Vipul Kumar, Singh, Dilbag, Agrawal, Apeksha
Format: Conference Proceeding
Language:English
Subjects:
Communication systems
Computational modeling
Convolutional Neural Network
Deep learning
fastText
InceptionV3
Long short-term memory
Measurement
Memory architecture
Recurrent Neural Network
Recurrent neural networks
Semantics
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Numerous methods are available to describe images by words and image captioning is a well-examined and researched topic. Further, the subject was much studied to work with English. Our model functions to generate descriptive explanations for general images in Telugu using Deep Learning techniques. Trained on images having Telugu captions, the features of an image are initially extracted using InceptionV3, a pre-trained Convolutional Neural Network model whereas the syntactic and semantic features are processed by using pre-trained fastText word embeddings of Telugu on a Recurrent Neural Network of Long short-term memory architecture. This concept can spread the benefit of well-developed models to native Telugu speakers.
ISSN:2575-7288
DOI:10.1109/ICACCS51430.2021.9441908