Skip to main content

ODNET: Optimized DenseNet for Indian food classification

2023, vol.15 , no.4, pp. 27-36

Article [2023-04-03]

Authors
Jigar Patel
Hardik Talsania
Kirit Modi
Abstract

The field of food image classification and recognition is gaining prominence in academic research, primarily driven by its increasing significance in the domains of medicine and healthcare. The application of food image classification has the potential to enhance overall food experiences in various ways. In this study, optimized DenseNet architecture proposed for transfer learning. The experimental findings demonstrate that the optimized DenseNet model, accuracy rate of Training is 98.7% and testing is 95.10%, surpassing the performance of alternative model MobileNetv3 in direct comparison. Accuracy of MobileNetV3 on Indian food image dataset is 98% on training and 92.39% testing. It shows best model for Indian food image dataset is optimized DenseNet and performance of the system surpasses state of the art methods.

Keywords

Deep Learning, Food Classification, DenseNet, MobileNet, CNN

DOI

https://doi.org/10.59035/FPBL3081

Download full article

Citation of this article:

Jigar Patel , Hardik Talsania , Kirit Modi . ODNET: Optimized DenseNet for Indian food classification. International Journal on Information Technologies and Security, vol.15 , no.4, 2023, pp. 27-36. https://doi.org/10.59035/FPBL3081