Plastic Waste Management through the Development of a Low Cost and Light Weight Deep Learning Based Reverse Vending Machine

The management of the abundant amount used plastic bottle waste is a major concern nowadays, because it contributor to landfills and overburdens processing facilities. Once disposed of, can take centuries break down, hence, recycling not only manages efficiently, but reduces environmental impact creates economic opportunities, as well. An incentive-based Reverse Vending machine (RVM) an effective way involve general public in waste. existing solutions are either very expensive, from computation cost perspective, or they lack robustness durability necessary for deployment. In this research, we have developed low-cost RVM collection, identification, classification, sorting bottles with addition reward-based user application. includes computing device running lightweight image algorithm, sensors, camera, self-designed mechanical arm. To support low RVM, MobileNet model has been trained using transfer learning. A dataset 10,983 pet images was collected camera installed inside classification model. results study show that achieved 99.2% testing accuracy, which better than bottle-classification approaches. Moreover, weight 12 Mb, fourteen six times less inceptionV3 Res-Net (Residual Neural Networks), respectively. Furthermore, costs fraction price, compared solutions. Two versions deployed at University more 6 months, collecting over 650 kg