Image Based Classification of Waste Material Using Convolution Neural Network

  • Chandaluru Priyanka, P. Sriramya


This paper describes Image based classification of the huge amount of waste which is thrown by people in civic areas. The workers can segregate the huge amounts of waste into different categories by hand is the old method which affects the hygienic diseases and this method will takes a long time to classify the waste in the municipal areas. So, this paper implements the classification of waste material based on the image. Sorting is a multi-stage process used to classify the different objects into their specific categories. The current classification process requires conveniences to sort the waste by hand and use a series of large screens to separate out more defined object. The main objective of this project is to find an easy approach to differentiate different objects and classify them and group them according to their categories. In this process we use Convolutional Neural Network (CNN), Tensor Flow and Transfer Learning for the prediction of existing models. The approach is to take picture of the objects, classify the images and process them into different categories. The domain which we applied is the Deep learning using Convolutional Neural Network (CNN) algorithm. It is defined by training the system with the large datasets which is used to identify the images and test set results will predict the accuracy results. The materials used for experimentation are - Cardboard Sheets, Plastic bottles, Glass items, Metals like tins, Sample papers like Whitepapers, Newspapers. This will bring an efficient time to sort the waste and it is the process of not getting any harm to anyone. By using this algorithm, it will predict the results more accurately.

Keywords: Waste material, convolutional neural network (CNN), Transfer Learning, Sorting, classification.

How to Cite
Chandaluru Priyanka, P. Sriramya. (2020). Image Based Classification of Waste Material Using Convolution Neural Network. International Journal of Advanced Science and Technology, 29(05), 2967 - 2975. Retrieved from