Performance Analysis of Image Prediction using Keras and Gradio: A Comparative Study

Heryanto, Yayan and Fauziah, - Performance Analysis of Image Prediction using Keras and Gradio: A Comparative Study. Performance Analysis of Image Prediction using Keras and Gradio: A Comparative Study. (Submitted)

[img] Text
Performance Analysis of Image Prediction using Keras and Gradio A Comparative Study_template IJACSA_revisi.doc
Download (322kB)

Abstract

This research titled "Performance Analysis of Image Prediction using Keras and Gradio: A Comparative Study" aims to analyze and compare the performance of image prediction using two specific frameworks, Keras and Gradio. The study evaluates the effectiveness and efficiency of both frameworks in accurately predicting images. Testing is conducted by training and evaluating deep learning models using a dataset of images collected from 10 different animal species, with each species having 5 sample images, resulting in a total of 50 animal images tested. The research also compares various architectures and optimization techniques to enhance the predictive capabilities of the models. Performance metrics considered in the study include accuracy and training time. The results show that using Gradio for image prediction yields faster processing times compared to Keras. The average processing time using Gradio is 1.2 seconds, while with Keras, it is 3.36 seconds. Furthermore, Gradio achieves a higher accuracy rate, with 360 out of 500 (72%) correct answers, whereas Keras only reaches 345 out of 500 (69%) correct answers. These findings demonstrate that Gradio performs better in terms of accuracy and processing efficiency compared to Keras in the task of image prediction for similar animal categories. The results of this research can provide valuable insights for researchers and practitioners in selecting the most suitable framework for image prediction projects involving similar animal species.

Item Type: Article
Subjects: T Technology > T Technology (General)
Divisions: Artikel > Informatika dan Sistem Informasi
Depositing User: Hesti Ari Wardani
Date Deposited: 21 Jul 2023 03:02
Last Modified: 21 Jul 2023 03:02
URI: http://repository.unas.ac.id/id/eprint/7860

Actions (login required)

View Item View Item
Unas Repository is powered by EPrints 3 which is developed by the School of Electronics and Computer Science at the University of Southampton. More information and software credits.