EUVIMEDEUVIMED entdecken
Crossref

Deep Learning‐Based Waste Image Classification and Web‐Based Environmental Awareness for Sustainable Waste Management

Mohammad Hashim, Tanaya Ganguly, Vishal Kumar Parida, Sumira Malik, Perumal Asaithambi

Environmental Quality Management · 2026-11 · 36 (4)

DOI: 10.1002/tqem.70509 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Environmental Quality Management, online erschienen am 10.11.2026. Verfasst von Mohammad Hashim und 4 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Rapid urbanization and improper waste segregation have increased the need for effective and sustainable waste‐management practices. Artificial intelligence (AI) and computer vision provide opportunities for automated waste identification and user‐oriented waste‐management support. This study presents EcoSort AI, an integrated deep‐learning‐based waste classification and sustainability‐awareness framework. An EfficientNetB0 model with ImageNet‐pretrained weights was trained to perform nine‐class multiclass classification of waste images covering wood, textiles, plastic, paper & cardboard, organic waste, glass, e‐waste, carton, and aluminium. The dataset contained 8346 images initially, of which 111 organic‐waste images were removed during data preparation, resulting in 8235 images for model development. The cleaned dataset of 8235 images was divided into training (5763 images), validation (1234 images), and held‐out test (1238 images) subsets using stratified sampling. The model achieved a test accuracy of 92.97%, with weighted precision, recall, and F1‐score of 92.99%, 92.97%, and 92.95%, respectively. The trained model was integrated into a web‐based EcoSort AI application that demonstrates image‐based waste identification and provides category‐specific sustainability profiles, waste‐management recommendations, environmental perspectives, and Sustainable Development Goal (SDG) mapping. The environmental‐impact component uses static, category‐level reference information and does not calculate an item‐specific carbon footprint or site‐specific life‐cycle emissions. The results demonstrate the classification performance and functional integration of the proposed framework, while independent external validation, real‐world user evaluation, and broader operational deployment remain areas for future research.

Zitieren

APA

Hashim, M., Ganguly, T., Parida, V. K., Malik, S., & Asaithambi, P. (2026). Deep Learning‐Based Waste Image Classification and Web‐Based Environmental Awareness for Sustainable Waste Management. Environmental Quality Management, 36(4). https://doi.org/10.1002/tqem.70509

Vancouver

Hashim M, Ganguly T, Parida VK, Malik S, Asaithambi P. Deep Learning‐Based Waste Image Classification and Web‐Based Environmental Awareness for Sustainable Waste Management. Environmental Quality Management. 2026;36(4). doi:10.1002/tqem.70509

BibTeX
@article{Hashim2026,
  author = {Hashim, Mohammad and Ganguly, Tanaya and Parida, Vishal Kumar and Malik, Sumira and Asaithambi, Perumal},
  title = {Deep Learning‐Based Waste Image Classification and Web‐Based Environmental Awareness for Sustainable Waste Management},
  journal = {Environmental Quality Management},
  year = {2026},
  volume = {36},
  number = {4},
  doi = {10.1002/tqem.70509}
}
Kostenlos starten →