Determining the fullness of garbage containers by deep learning
[ X ]
Tarih
2023
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Elsevier Ltd
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
An essential point in waste management, which is a matter of great importance for the environment and nature, is waste collection from temporary storage points. Since the garbage collection process is generally time-related, sometimes the garbage containers overflow or empty. Intelligent services are being developed for issues related to the cleanliness of the streets through cameras and specially designed monitoring tools. This study has investigated whether deep learning can determine if the garbage containers are full or not based on the camera images. For this purpose, experiments were carried out for automatic classification processes by applying DenseNet-169, EfficientNet-B3, MobileNetV3-Large, and VGG19-Bn deep learning algorithms on the CDCM dataset, which contains images of trash cans or containers labeled as clean and dirty. With a 94.931% accuracy rate, it has been found that an intelligent system can be used successfully in smart cities to determine the status of garbage and garbage containers on the streets and inform the authorities. © 2023 Elsevier Ltd
Açıklama
Anahtar Kelimeler
Deep learning, Garbage container, Smart city, Street cleanliness
Kaynak
Expert Systems with Applications
WoS Q Değeri
Scopus Q Değeri
Q1
Cilt
217
