Research Article
An Attention-Enhanced Framework for Automated Fruit Ripeness and Quality Assessment
@INPROCEEDINGS{10.1007/978-3-032-44094-5_2, author={Giang Vu Van and Tue Dinh Huynh Tue and Bay Vo and Hoang Thien Van}, title={An Attention-Enhanced Framework for Automated Fruit Ripeness and Quality Assessment}, proceedings={Industrial Networks and Intelligent Systems. 12th EAI International Conference, INISCOM 2026, Da Nang, Vietnam, February 26--27, 2026, Proceedings}, proceedings_a={INISCOM}, year={2026}, month={11}, keywords={Convolutional Neural Networks Convolutional Block Attention Module Fruit Quality DenseNet Post Harvest Agriculture Tomato Chili MultiFruit Quality}, doi={10.1007/978-3-032-44094-5_2} }- Giang Vu Van
Tue Dinh Huynh Tue
Bay Vo
Hoang Thien Van
Year: 2026
An Attention-Enhanced Framework for Automated Fruit Ripeness and Quality Assessment
INISCOM
Springer
DOI: 10.1007/978-3-032-44094-5_2
Abstract
In the agricultural industry, classifying fruit ripeness and quality play important role for proper harvesting, effective storage, food safety, and marketability. Deep learning has a lot of potential to solve these problems, while manual inspection is often subjective and not very effective. Even though a lot of research has been done, traditional CNN models still have problems with redundant feature extraction and not being able to capture fine-grained surface details of agricultural products. This study proposes DenseNetCBAM model for fruit quality classification. This model integrates channel and spatial attention modules into DenseNet121 to improve feature representation and focus on quality-degraded areas. Experiments on the evaluation datasets VegNet and FruitNet show that DenseNetCBAM achieves an accuracy of 98.15% - 99.78% and a higher F1 than other state-of-the-art models. The model reduces misclassification in difficult classes, keeps training and inference time competitive and has a compact size. This study demonstrates the effectiveness of the integrated model in classifying the ripness and quality of fruits and encourages further innovation for practical agricultural applications.