Transfer learning with deep models for small food datasets

Longzheng Cai; Longmei Tang; Shuyun Lim

doi:10.1117/12.2686482

10 August 2023 Transfer learning with deep models for small food datasets

Longzheng Cai, Longmei Tang, Shuyun Lim

Proceedings Volume 12759, International Conference on Automation Control, Algorithm, and Intelligent Bionics (ACAIB 2023); 127591E (2023) https://doi.org/10.1117/12.2686482
Event: 2023 3rd International Conference on Automation Control, Algorithm and Intelligent Bionics (ACAIB 2023), 2023, Xiamen, China

Abstract

Deep Convolutional Neural Networks (DCNNs) have been used in food image classification, segmentation, gradient identification and many other applications. There are many small segments in food industry, and it is very expensive and time-consuming to build a large training dataset for each of them. Transfer learning can play a role in this aspect by reusing knowledge learned from large datasets, like ImageNet, to target tasks where only small datasets are available. In this work, we conduct an empirical study on the transfer learning capability of DCNNs pre-trained with ImageNet on four small food datasets. We find that models perform better on ImageNet dataset also have better transfer performance for small target datasets; In two transfer learning approaches, fine-turning performs better than feature extraction; Data augmentation helps in transfer learning.

Citation Download Citation

Longzheng Cai, Longmei Tang, and Shuyun Lim "Transfer learning with deep models for small food datasets", Proc. SPIE 12759, International Conference on Automation Control, Algorithm, and Intelligent Bionics (ACAIB 2023), 127591E (10 August 2023); https://doi.org/10.1117/12.2686482

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available