Image-Based Waste Segregation Using Machine Learning

Image-Based Waste Segregation Using Machine Learning

Nazhat Sultana S 1,Mrs Roopa R2

[1]Student, Department of MCA, BIET, Davanagere

[2]Assistant Professor, Department of MCA, BIET, Davanagere

ABSTRACT

While waste disposal is essential to preserving biodiversity, adequate sorting of waste is still difficult and tedious. Due to the 7,000 tons of trash produced every day in the region, source classification of garbage and degradation is necessary for safer methods of trash disposal. With the use of neural networks made up of convolutions (CNNs), separating trash can now be automated with 92% accuracy. This is achieved even in noisy or variable-angle photos when identifying items such as glass, metal, cardboard, plastic, paper, and general rubbish. This technique is successful in sorting rubbish in the actual world, as seen by its extreme precision. India releases 62 million tons of garbage every year, of which approximately 5.6 million tons are made of plastic items. Plastic recycling is at 60%, but separating is still a major problem since it doesn't happen at the point of collection, requiring a lot of physical effort and putting employee well-being in danger. By using a CNN-based image classifier, this study seeks to decrease human involvement; With reference to the United States, which produced 264.4 million tons of municipal garbage in 2017, the report emphasizes the revolutionary potential of artificial intelligence and robots in garbage disposal.

Keywords: Machine learning, Convolutional Neural Networks (CNNs), Waste disposal, Image classification, VGG, ResNet, AlexNet.