Indian Food Image Recognition using a Deep Learning Approach
E. Emerson Nithiyaraj1, S. Rajaseela2
1E. Emerson Nithiyaraj, Research Scholar, Department of ECE, Mepco Schlenk Engineering College, Sivakasi (Tamil Nadu), India.
2S. Rajaseela, PG Student, Department of ECE, Pandian Saraswathi Yadav Engineering College, Sivaganga (Tamil Nadu), India.
Manuscript received on 12 November 2021 | Revised Manuscript received on 25 November 2021 | Manuscript Accepted on 15 December 2021 | Manuscript published on 30 December 2021 | PP: 1-5 | Volume-1 Issue-1 December 2021 | Retrieval Number: 100.1/ijfe.A1001121121
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Published by Lattice Science Publication (LSP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In today’s scenario, food image recognition is an interesting and useful application of visual object recognition. Convolutional Neural Network (CNN) is the best deep learning architecture for image classification tasks since it automatically learns the representations from the input images. Due to unlikeness and varieties of food available across the country, food recognition becomes very challenging. In this paper, a deep learning algorithm is proposed to recognize and classify 21 Indian food image categories. A transfer learning approach using Alex Net is developed for this task. For the experimentation, the dataset India-Food-21-Categories-Small is used from Kaggle and the Alex Net architecture is fine tuned for this application. Since the dataset has only limited amount of images, the available dataset is augmented to enhance the system’s performance. The proposed CNN architecture results in an accuracy of 96.6% while trained for just 5 epochs.
Keywords: Deep Learning, Convolutional Neural Network (CNN), Indian Food Image Recognition, Data Augmentation.
Scope of the Article: Indian Food Image Recognition