Comparing Convolutional Neural Networks for Waste Image Classification with Visual Explanation and Label Audit
DOI:
https://doi.org/10.47709/brilliance.v6i4.9844Keywords:
Convolutional Neural Network, explainable artificial intelligence, Label noise, MobileNetV2, Transfer learning, Waste classificationAbstract
Automated waste classification from images can support smart sorting bins, but the choice of a convolutional neural network for low-cost hardware is rarely evaluated together with the quality of the training labels. This study aims to compare three pretrained architectures, namely ResNet18, MobileNetV2, and EfficientNet-B0, for classifying waste images into recyclable, electronic, and organic classes, and to examine why their errors occur. The models were trained through transfer learning under an identical protocol on 26,527 images from the Satria Data 2026 Big Data Challenge, using class-weighted cross-entropy and early stopping, and were evaluated on a stratified validation set of 2,651 images in terms of accuracy, macro F1-score, parameter count, model size, and inference time. Gradient-weighted class activation mapping was applied as an explainable artificial intelligence technique, and the 40 most confident misclassifications were audited manually to estimate label noise. MobileNetV2 achieved the highest accuracy (92.68%) and macro F1-score (0.9239) while also being the smallest model (2.23 million parameters, 8.71 megabytes) and the fastest (16.78 milliseconds per image on a laptop processor). All three architectures confused recyclable and organic waste most often. The visual explanations showed a reliance on surface texture, and the audit attributed 30% of the confident errors to probable label noise, mostly organic images predicted as recyclable. Lightweight networks are therefore a sound starting point for smart bins, and label auditing should accompany accuracy reporting in waste classification studies.
References
Alk?l?nç, A., Okay, F. Y., Kök, ?., & Özdemir, S. (2026). Deep ensemble learning model for waste classification systems. Sustainability, 18(1), 24. https://doi.org/10.3390/su18010024
Altikat, A., Gulbe, A., & Altikat, S. (2022). Intelligent solid waste classification using deep convolutional neural networks. International Journal of Environmental Science and Technology, 19(3), 1285–1292. https://doi.org/10.1007/s13762-021-03179-4
Aprisia Bahagia, G., & Akbar, M. (2024). Klasifikasi sampah organik dan anorganik menggunakan metode convolutional neural network (CNN) [Classification of organic and inorganic waste using the convolutional neural network method]. JATI (Jurnal Mahasiswa Teknik Informatika), 8(5), 10349–10355. https://doi.org/10.36040/jati.v8i5.11038
Chhabra, M., Sharan, B., Elbarachi, M., & Kumar, M. (2024). Intelligent waste classification approach based on improved multi-layered convolutional neural network. Multimedia Tools and Applications, 83(36), 84095–84120. https://doi.org/10.1007/s11042-024-18939-w
Girsang, A. S., Saputra, A. D., & Yanrie, V. (2023). Performance comparison between VGG16 and Inception V3 for organic waste and recyclable waste classification. International Journal of Intelligent Systems and Applications in Engineering, 11(2), 557–563.
Goswami, M., Sanil, V., Choudhry, A., Srinivasan, A., Udompanyawit, C., & Dubrawski, A. (2023). AQuA: A benchmarking tool for label quality assessment. arXiv. https://doi.org/10.48550/arXiv.2306.09467
Gürcan, F., & Soylu, A. (2026). Automated waste classification for smart recycling: A multi-class CNN approach with transfer learning and pre-trained models. Environmental Technology & Innovation, 41, 104673. https://doi.org/10.1016/j.eti.2025.104673
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (pp. 770–778). Las Vegas, NV, USA: IEEE. https://doi.org/10.1109/CVPR.2016.90
Ibnul Rasidi, A., Pasaribu, Y. A. H., Ziqri, A., & Adhinata, F. D. (2022). Klasifikasi sampah organik dan non-organik menggunakan convolutional neural network [Classification of organic and non-organic waste using a convolutional neural network]. Jurnal Teknik Informatika Dan Sistem Informasi, 8(1). https://doi.org/10.28932/jutisi.v8i1.4314
Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv. https://doi.org/10.48550/ARXIV.1412.6980
Kumala, R. A., Sari, C. A., & Rachmawanto, E. H. (2025). A comparison of MobileNetV2 and VGG16 architectures with transfer learning for multi-class image-based waste classification. Journal of Applied Informatics and Computing, 9(4), 1610–1624. https://doi.org/10.30871/jaic.v9i4.9958
Liu, W., Ouyang, H., Liu, Q., Cai, S., Wang, C., Xie, J., & Hu, W. (2022). Image recognition for garbage classification based on transfer learning and model fusion. Mathematical Problems in Engineering, 2022, 1–12. https://doi.org/10.1155/2022/4793555
Manik, F. Y., Nainggolan, P. I., Harumy, T. H. F., Ginting, D. S. B., Maharani, A., Ramadayanti, H., … Harifin, M. P. (2026). Explainable CNN-based multiclass household waste classification using Grad-CAM for smart waste management. International Journal of Advanced Computer Science and Applications, 17(1), 580–589. https://doi.org/10.14569/IJACSA.2026.0170155
Mehedi, M. H. K., Arafin, I., Hasan, M., Rahman, F., Tasin, R., & Rasel, A. A. (2023). A transfer learning approach for efficient classification of waste materials. In 2023 IEEE 13th Annual Computing and Communication Workshop and Conference (CCWC), (pp. 636–640). Las Vegas, NV, USA: IEEE. https://doi.org/10.1109/CCWC57344.2023.10099127
Muslihati, M., Sahibu, S., & Taufik, I. (2024). Implementasi algoritma convolutional neural network untuk klasifikasi jenis sampah organik dan non organik [Implementation of the convolutional neural network algorithm for classifying types of organic and non-organic waste]. MALCOM: Indonesian Journal of Machine Learning and Computer Science, 4(3), 840–852. https://doi.org/10.57152/malcom.v4i3.1346
Nnamoko, N., Barrowclough, J., & Procter, J. (2022). Solid waste image classification using deep convolutional neural network. Infrastructures, 7(4), 47. https://doi.org/10.3390/infrastructures7040047
Northcutt, C. G., Athalye, A., & Mueller, J. (2021). Pervasive label errors in test sets destabilize machine learning benchmarks. In Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks. Retrieved from https://openreview.net/forum?id=XccDXrDNLek
Northcutt, C. G., Jiang, L., & Chuang, I. L. (2021). Confident learning: Estimating uncertainty in dataset labels. Journal of Artificial Intelligence Research, 70, 1373–1411. https://doi.org/10.1613/jair.1.12125
Oktayaessofa, E., Sari, C. A., Rachmawanto, E. H., & Yaacob, N. M. (2024). Classification of organic and non-organic waste with CNN-MobileNet-V2. Jurnal Teknik Informatika (JUTIF), 5(4), 1173–1180. https://doi.org/10.52436/1.jutif.2024.5.4.2165
Poudel, S., & Poudyal, P. (2022). Classification of waste materials using CNN based on transfer learning. IN Proceedings of the 14th Annual Meeting of the Forum for Information Retrieval Evaluation, (pp. 29–33). Kolkata, India: ACM. https://doi.org/10.1145/3574318.3574345
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L.-C. (2018). MobileNetV2: Inverted residuals and linear bottlenecks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, (pp. 4510–4520). Salt Lake City, UT: IEEE. https://doi.org/10.1109/CVPR.2018.00474
Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2017). Grad-CAM: Visual explanations from deep networks via gradient-based localization. In 2017 IEEE International Conference on Computer Vision (ICCV), (pp. 618–626). Venice: IEEE. https://doi.org/10.1109/ICCV.2017.74
Shetty, S., Kallianpur, S., Fernandes, R., Rodrigues, A. P., & Padmanabha, V. (2025). ECO-HYBRID: Sustainable waste classification using transfer learning with hybrid and enhanced CNN models. Sustainability, 17(19), 8761. https://doi.org/10.3390/su17198761
Shreya, M., Nimal Yughan, V., Katyal, J., & Ramesh, R. (2023). Technical solutions for waste classification and management: A mini-review. Waste Management & Research: The Journal for a Sustainable Circular Economy, 41(4), 801–815. https://doi.org/10.1177/0734242X221135262
Song, H., Kim, M., Park, D., Shin, Y., & Lee, J.-G. (2023). Learning from noisy labels with deep neural networks: A survey. IEEE Transactions on Neural Networks and Learning Systems, 34(11), 8135–8153. https://doi.org/10.1109/TNNLS.2022.3152527
Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. Proceedings of the 36th International Conference on Machine Learning, 6105–6114. PMLR. Retrieved from https://proceedings.mlr.press/v97/tan19a.html
Yong, L., Ma, L., Sun, D., & Du, L. (2023). Application of MobileNetV2 to waste classification. PLOS ONE, 18(3), e0282336. https://doi.org/10.1371/journal.pone.0282336
Zhang, Y., Kang, B., Hooi, B., Yan, S., & Feng, J. (2023). Deep long-tailed learning: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(9), 10795–10816. https://doi.org/10.1109/TPAMI.2023.3268118
Zhuang, F., Qi, Z., Duan, K., Xi, D., Zhu, Y., Zhu, H., … He, Q. (2021). A comprehensive survey on transfer learning. In Proceedings of the IEEE, 109(1), (pp. 43–76). https://doi.org/10.1109/JPROC.2020.3004555
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Teuku Muhammad Mirza Keumala, Mohd. Iqbal Muttaqin, Nur Aynun Siregar
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
