•  
  •  
 

ORCID

Anupriya: 0000-0002-2245-4092 Ratish : 0000-0003-1020-558X

Subject Area

Computer Science

Article Type

Original Study

Abstract

Efficient waste segregation into biodegradable and non-biodegradable categories is a crucial requirement for sustainable smart city development. While Convolutional Neural Network (CNN)-based approaches have demonstrated promising performance, their high architectural complexity and limited adaptability often restrict real-time deployment. This paper proposes BnBCNet, a lightweight waste classification framework that reformulates image-based waste segregation as a sequential learning problem using Gated Recurrent Units (GRUs). To further enhance classification performance, Grey Wolf Optimization (GWO) is employed for hyperparameter optimization, while Adam optimizer is used for network weight updates. Experiments conducted on a publicly available waste segregation image dataset demonstrate that the proposed BnBCNet achieves a classification accuracy of 98.04%, outperforming baseline GRU models and several state-of-the-art CNN-based methods. The proposed architecture also achieves significant loss reduction and improved correct classification rate, confirming its effectiveness and scalability for real-time smart waste management systems.

Keywords

Waste Classification, Gated Recurrent Units (GRU), Grey Wolf Optimization (GWO), Sequential Pattern Learning, Smart Cities

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Download

Share

COinS