A Low-Cost Solution for Automatic Plastic Segregation
Meher Madhu Dharmana1, Aiswarya M. S.2
1Meher Madhu Dharmana*, Department of Electrical and Electronics Engineering, Amrita Vishwa Vidyapeetham, Amritapuri, India.
2Aiswarya M S, Department of Electrical and Electronics Engineering, Amrita Vishwa Vidyapeetham, Amritapuri, India.
Manuscript received on February 01, 2020. | Revised Manuscript received on February 05, 2020. | Manuscript published on February 30, 2020. | PP: 784-788 | Volume-9 Issue-3, February, 2020. | Retrieval Number: C5275029320/2020©BEIESP | DOI: 10.35940/ijeat.C5275.029320
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Solid waste management is a universal issue that matters to every single person in the world. The solid waste management system is fundamentally labor intensive with very little collection efficiency. The available automatic plastic segregation techniques are based on thermal imaging and electrostatic properties of materials- these methods are expensive for governments to invest upon, and also to maintain in landfills. In this paper, artificial intelligence techniques are exploited to recognize the sounds of plastics from that of other materials by designing suitable mechanism to produce sound from debris during segregation, the segregation process can be automated with relatively low-cost electronics like System on Chips and audio sensors. With 30,000 recorded samples of noisy plastic and non-plastic material sounds, ANN is trained and was able to successfully detect plastics with 93.5% accuracy in real time. Algorithms were developed in python and real time testing was done on SoC with a mic, which affirms that the proposed method is cost effective when compared to techniques involving image processing, thermal imaging and near infrared spectroscopy.
Keywords: Machine learning, features, classifier, audio signal.