Exploring ResNet-18 Estimation Design through Multiple Implementation Iterations and Techniques in Legacy Databases

Journal of Robotics and Control (JRC), Год журнала: 2023, Номер 4(5), С. 650 - 661

Опубликована: Сен. 21, 2023

In a rapidly evolving landscape where automated systems and database applications are increasingly crucial, there is pressing need for precise efficient object recognition methods. This study contributes to this burgeoning field by examining the ResNet-18 architecture, proven deep learning model, in context of fruit image classification. The research employs an elaborate experimental setup featuring diverse dataset that includes Rambutan, Mango, Santol, Mangosteen, Guava. efficacy single versus multiple models compared, shedding light on their relative classification accuracy. A unique aspect establishment 90% decision threshold, introduced mitigate risk incorrect Our statistical analysis reveals significant performance advantage over models, with average improvement margin 15%. finding substantiates study’s central hypothesis. implemented threshold determined play pivotal role augmenting system’s overall accuracy minimizing false positives. However, it’s worth noting increased computational complexity associated deploying necessitates further scrutiny. sum, provides nuanced evaluation realm classification, emphasizing utility practical, real-world applications. opens avenues future exploration refining these methodologies investigating applicability broader tasks.

Язык: Английский

Fruit Zone : Media Pembelajaran Interaktif Pengenalan Buah Anak Kelompok Belajar Menggunakan ResNet18

Faktor Exacta, Год журнала: 2024, Номер 17(1)

Опубликована: Май 2, 2024

Learning media is very important in supporting learning activities early childhood. Limited and methods that are still centered on the ability experience of teachers an obstacle to improving at Pos Alamanda 105 Jumerto, Jember. An interactive, cheap, easy accessible needed improve students' abilities, especially fruit recognition using both Indonesian English. The solution, researchers used Deep method for interactive introduction Convolutional Neural Network with Resnet18 architecture. This research uses 21 types popular fruits unique equipped voice features total data 2100 images a rate 0.0002 maximum epoch 100 wereable classify accuracy 96% (system training) 95% testing).

Язык: Английский