Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Health Care Waste Classification Using Deep Learning Aligned with Nepal's Bin Color Guidelines

Suman Kunwar, Prabesh Rai

TL;DR

The paper tackles the challenge of health care waste segregation in Nepal by benchmarking multiple deep learning models on a combined HCW dataset aligned with the National HCW color-coding standards. Using 5-fold stratified cross-validation, it compares YOLOv5-s, YOLOv8-n, EfficientNet-B0, ResNeXt-50, and MobileNetV3-S, finding YOLOv5-s achieves the highest accuracy of 95.06%. Inference speed and model size influence practical deployment, with YOLOv8-n approaching similar accuracy but slower, and EfficientNet-B0 delivering strong accuracy at the cost of longer inference; ResNeXt-50 performs considerably worse. The authors deployed YOLOv5-s to a web interface with Nepal color-bin mapping and discuss limitations (missing some categories, non-Nepali data) and future work focusing on data collection and localization for Nepal-specific HCW sorting.

Abstract

The increasing number of Health Care facilities in Nepal has added up the challenges on managing health care waste (HCW). Improper segregation and disposal of HCW leads to contamination, spreading of infectious diseases and risk for waste handlers. This study benchmarks the state of the art waste classification models: ResNeXt-50, EfficientNet-B0, MobileNetV3-S, YOLOv8-n and YOLOv5-s using stratified 5-fold cross-validation technique on combined HCW data. YOLOv5-s achieved the highest accuracy (95.06%) but fell short with the YOLOv8-n model in inference speed with few milliseconds. The EfficientNet-B0 showed promising results of 93.22% accuracy but took the highest inference time. Following a repetitive ANOVA test to confirm the statistical significance, the best performing model (YOLOv5-s) was deployed to the web with bin color mapped using Nepal's HCW management standards. Further work is suggested to address data limitation and ensure localized context.

Health Care Waste Classification Using Deep Learning Aligned with Nepal's Bin Color Guidelines

TL;DR

The paper tackles the challenge of health care waste segregation in Nepal by benchmarking multiple deep learning models on a combined HCW dataset aligned with the National HCW color-coding standards. Using 5-fold stratified cross-validation, it compares YOLOv5-s, YOLOv8-n, EfficientNet-B0, ResNeXt-50, and MobileNetV3-S, finding YOLOv5-s achieves the highest accuracy of 95.06%. Inference speed and model size influence practical deployment, with YOLOv8-n approaching similar accuracy but slower, and EfficientNet-B0 delivering strong accuracy at the cost of longer inference; ResNeXt-50 performs considerably worse. The authors deployed YOLOv5-s to a web interface with Nepal color-bin mapping and discuss limitations (missing some categories, non-Nepali data) and future work focusing on data collection and localization for Nepal-specific HCW sorting.

Abstract

The increasing number of Health Care facilities in Nepal has added up the challenges on managing health care waste (HCW). Improper segregation and disposal of HCW leads to contamination, spreading of infectious diseases and risk for waste handlers. This study benchmarks the state of the art waste classification models: ResNeXt-50, EfficientNet-B0, MobileNetV3-S, YOLOv8-n and YOLOv5-s using stratified 5-fold cross-validation technique on combined HCW data. YOLOv5-s achieved the highest accuracy (95.06%) but fell short with the YOLOv8-n model in inference speed with few milliseconds. The EfficientNet-B0 showed promising results of 93.22% accuracy but took the highest inference time. Following a repetitive ANOVA test to confirm the statistical significance, the best performing model (YOLOv5-s) was deployed to the web with bin color mapped using Nepal's HCW management standards. Further work is suggested to address data limitation and ensure localized context.

Paper Structure

This paper contains 5 sections, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Dataset and Preprocessing
  4. Model Training
  5. Results and Discussion

Figures (5)

  • Figure 1: Color codes for waste segregation
  • Figure 2: Waste bins at Bhaktapur Cancer Hospital, Nepal
  • Figure 3: Training loss and accuracy curves for each fold in 5-fold cross-validation of EfficientNet-B0 model
  • Figure 4: Model Performance Comparison Across Metrics using 5 fold
  • Figure 5: HCW Detection sample app with bin color based on Nepal's guidelines