Lotus species classification using transfer learning based on VGG16, ResNet152V2, and MobileNetV2

Nachirat Rachburee, Wattana Punlumjeak

Abstract


Technology has played an increasingly important role in daily life. Especially, technology in object classification that comes in to make human life more comfortable as well as to help people of all ages learning unlimited in anywhere, anytime. Lotus Museum located in Rajamangala University of Technology Thanyaburi (RMUTT) that is open to the general public to learn as well as to cultivate awareness for propagation and result in future preservation. In this paper, we proposed lotus species classification with three pre-trained weights in the ImageNet dataset: visual geometry group (VGG16), residual neural network (ResNet152V2), and MobileNetV2. Fine-turning is used in the last layer after retrained with the custom data we provided. The experimental result shows the accuracy of VGG16, ResNet152V2, and MobileNetV2 are 98.5%, 98.0%, and 99.5% respectively. Therefore, MobileNetV2 not only gives the best accuracy than others but also uses the lowest parameters which are effective in computation time and proper to mobile devices. The proposed research paper on lotus classification base on transfer learning is an effective way to encourage and support people to learn without limitations.

Keywords


Classification; Convolution neural network; Lotus species classification; Pre-trained weight; Transfer learning;

Full Text:

PDF


DOI: http://doi.org/10.11591/ijai.v11.i4.pp%25p

Refbacks

  • There are currently no refbacks.


View IJAI Stats

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.