<![CDATA[According to Law Number 18 of 2008, the increase in the volume, type and characteristics of waste is caused by population growth and changes in people's consumption patterns. Regulation of the Minister of Environment and Forestry of the Republic of Indonesia Number P.10/MENLHK/SETJEN/PLB.0/4/2018, concerning the handling of household waste and similar waste, it is necessary to classify it. In general, waste is divided into three types, namely Inorganic, Organic and B3. Many people already know the types of waste but are still not sure about sorting waste properly. It is necessary to recognize objects to design a household waste classification system to make it easier to sort waste more optimally. This system involves Deep Learning and in it there is an algorithm that is able to classify objects significantly, namely the Convolutional Neural Network (CNN) algorithm. This research uses the MobileNet architecture, which is a CNN architecture that has fast and accurate computing time. The dense layer contains 1 layer with 900 neurons. There are 5952 trash dataset images that will be used as training data and 1489 testing data taken from the kaggle.com dataset. The classification process for training data using the MobileNet architecture produces an average score for each class, namely an accuracy value of 88%, precision of 89%, recall of 88%, and F1 of 87%. Meanwhile, the results of the training model applied to the testing data produced an accuracy of 86%. Thus, the results of this experiment have quite good accuracy considering that household waste has various types and shapes.]]>
Copyrights © 2024
