Authors :
Ashik Mohammed Sali; Harish Thampy; Immanuel S Vadakedam
Volume/Issue :
Volume 5 - 2020, Issue 11 - November
Google Scholar :
http://bitly.ws/9nMw
Scribd :
https://bit.ly/37nbIyL
Abstract :
Waste management is an important issue in the
current scenario. Sorting of waste into different
categorizes is one of the most important and tedious step
in waste management.
Normally this is a manual (hand-picking) process
which has its own cons and hence the need for an
automated and efficient system to manage waste arises.
Through this paper we propose an intelligent waste
classification system, developed using a deep pre-trained
(Xception) Convolutional Neural Network model that can
classify solid wastes such as glass, metal, paper and plastic
etc. on stand-alone edge devices. This system can be
further deployed into a real time embedded system by
adding a mechanism to physically separate wastes. The
proposed system is trained on an open source dataset
available online and is able to achieve a test accuracy of
92% on the dataset. Thus the system could make the
separation process faster and intelligent without or
reducing human involvement.
Keywords :
Convolutional Neural Networks, Machine Learning, Deep Learning, Xception model, Dataset, Training, Sorting, Edge computing.
Waste management is an important issue in the
current scenario. Sorting of waste into different
categorizes is one of the most important and tedious step
in waste management.
Normally this is a manual (hand-picking) process
which has its own cons and hence the need for an
automated and efficient system to manage waste arises.
Through this paper we propose an intelligent waste
classification system, developed using a deep pre-trained
(Xception) Convolutional Neural Network model that can
classify solid wastes such as glass, metal, paper and plastic
etc. on stand-alone edge devices. This system can be
further deployed into a real time embedded system by
adding a mechanism to physically separate wastes. The
proposed system is trained on an open source dataset
available online and is able to achieve a test accuracy of
92% on the dataset. Thus the system could make the
separation process faster and intelligent without or
reducing human involvement.
Keywords :
Convolutional Neural Networks, Machine Learning, Deep Learning, Xception model, Dataset, Training, Sorting, Edge computing.
