Metaheuristics with Deep Transfer Learning Enabled Detection and classification model for industrial waste management

Area of Science:

• Environmental Science
• Computer Science
• Engineering

Background:

• Industrialization generates substantial solid and liquid waste, posing environmental and health risks due to inadequate management and disposal.
• Current waste segregation relies on manual methods, which are inefficient and hinder effective industrial waste management (IWM).
• There is a need for automated, smart techniques to accurately classify and manage industrial waste.

Purpose of the Study:

• To introduce a novel Metaheuristics with Deep Transfer Learning Enabled Detection and Classification Methods for Industrial Waste Management (MDTLDC-IWM) model.
• To enhance the efficiency and accuracy of industrial waste identification and classification.
• To provide an automated solution for waste segregation and management.

Main Methods:

• The MDTLDC-IWM model employs a two-phase approach: waste object recognition and classification.
• Waste object recognition utilizes the YOLO-v5 object detector optimized with the Harris Hawks Optimization (HHO) algorithm.
• Waste object classification is performed using a stacked sparse auto encoder (SSAE) model, optimized by the Aquila Optimization Algorithm (AOA).

Main Results:

• The MDTLDC-IWM model achieved a high precision of 96.84% and an F-score of 96.71%.
• Experimental validation on a benchmark dataset confirmed the model's effectiveness.
• Comparative analysis demonstrated superior performance over existing state-of-the-art methods.

Conclusions:

• The MDTLDC-IWM model offers a significant advancement in automated industrial waste management.
• The integration of metaheuristics and deep transfer learning provides an accurate and efficient waste classification system.
• This approach contributes to better environmental protection and public health by improving waste segregation processes.