| dc.description.abstract | Waste management always is the problem of global, and image recognition based on deep
learning has been designed to enhance the accuracy and reduce labor cost. In the topic of
waste image classification, there several research have proposed many method to increase
the accuracy on many dataset. In this study, the goal is developed an automated waste
image classification system using convolutional neural networks (CNNs) and building a
new dataset based on the categories of Vietnam. Image recognition and computer vision
methods will be explored for accurate waste classification. CNN model, which is
DenseNet121or ResNet18 will be trained on new dataset for feature extraction from waste
images. These extracted features will be used to train task-specific classifiers for waste
categories relevant to the application context. Furthermore, the result’s analysis will show
the performance of model or the possibility of dataset. This research will demonstrate a
feasible solution for real-world waste classification challenges by leveraging deep learning.
The techniques can potentially be extended to large-scale waste management systems to
enable automated sorting for recycling. | en_US |
