Inovasi Hybrid Neural Network Algoritma Single Shot Multibox Detector (SSD) dan ResNet -18 Identifikasi dan Klasifikasi Sampah Pesisir (Studi Kasus Pantai Olo Medan)

Abstract

Coastal waste is a serious environmental problem in Indonesia, including at Olo Beach Medan. Inefficient coastal waste management can lead to marine pollution, ecosystem damage, and public health problems. Therefore, an effective method for identifying and classifying coastal waste is needed. This study proposes a hybrid neural network innovation using the Single Shot Multibox Detector (SSD) and ResNet-18 algorithms for coastal waste identification and classification. The SSD algorithm is used to detect the location of coastal waste in images, while ResNet-18 is used to classify the type of coastal waste. This hybrid model is trained using a dataset of coastal waste images from Olo Beach Medan. The results show that the hybrid SSD and ResNet-18 model has high accuracy in identifying and classifying coastal waste. Detection accuracy reaches 95%, while classification accuracy reaches 90%. The model can detect and classify various types of coastal waste, such as plastic waste, organic waste, and wood waste. The hybrid neural network innovation using the SSD and ResNet-18 algorithms has the potential to be an effective method for coastal waste identification and classification. This model can be used to assist in more optimal coastal waste management at Olo Beach Medan and other coastal areas.