Garbage Classification Using CNN Architecture ShuffleNet v2

The increase in population has a serious impact on aspects of an area or city, one of which is the overflow of waste volume. According to data from the National Waste Management Information System (SIPSN) quoted on the official SIPSN website, in 2021 the amount of Indonesia's waste piles will be 28.57 million tons/year and 78.27 million tons/day. Reduction of waste by 15.62% or 4.4 million tons from the previous year, but waste management in Indonesia can only be managed as much as 64.52% or 18.43 million tons/year. This is due to the lack of knowledge and awareness of the surrounding community in classifying waste when disposing of waste that is not in accordance with its type. The practice of garbage disposal is sometimes still confused about the type of waste category they have by reason of the large number of materials that make up the waste. Based on these problems, a study was conducted on the classification of types of waste to determine whether the category can be recycled. This study will classify the types of waste using 6601 waste image data with the division of waste types as many as 4 classes, namely plastic waste, organic paper and others using ShuffleNet v2. And it can be concluded that the ShuffleNet v2 architecture is able to classify garbage images with four data groups that have been selected from different sources on the Kaggle website with recall, precision, and f1-score values on ShuffleNet v2 with a number of epochs are 50, are 97.32%, 96.79%, and 97.03%.