Abstract

This paper introduces ReLeaf, an intelligent waste classification system that integrates deep learning and mobile application technologies to promote environmentally responsible waste disposal. The approach employs the MobileNetV2 architecture with transfer learning and fine-tuning to classify waste into six categories: cardboard, glass, metal, paper, plastic, and trash. Various data augmentation techniques were systematically evaluated to enhance model generalization and mitigate overfitting. The model was trained on a combination of public datasets and achieved a test accuracy of 93.96%. The trained model was deployed within a Flutterbased mobile application using TensorFlow Lite, enabling realtime waste recognition on mobile devices. Additionally, a cloudbased large language model (GPT-4.1-mini) was incorporated to provide recycling guidance and user assistance through natural language interaction. Experimental results indicate that the proposed system offers an efficient, practical solution for intelligent waste classification and recycling support in real-world settings.

References

• Aulia Afifah, Endah Ratna Arumi, Maimunah Maimunah, Setiya Nugroho, et al. Comparative evaluation of preprocessing methods for mobilenetv1 and v2 in waste classification. Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi), 9(3):194–202, 2025.
• Gulzar Ahmad, Fizza Muhammad Aleem, Tahir Alyas, Qaiser Abbas, Waqas Nawaz, Taher M Ghazal, Abdul Aziz, Saira Aleem, Nadia Tabassum, and Aidarus Mohamed Ibrahim. Intelligent waste sorting for urban sustainability using deep learning. Scientific reports, 15(1):27078, 2025.
• Iman Ameer Ahmad, Ahmed Mudheher Hasan, and Amjad Jaleel Humaidi. Development of a memory-efficient and computationally cost-effective cnn for smart waste classification. Journal of Engineering Research, 2025.
• Imam Husni Al Amin, Dwiki Vio Setyadarma, and Setyawan Wibisono. Integration of the faster r-cnn algorithm for waste detection in an android application. Revue d’Intelligence Artificielle, 37(6):1407, 2023.
• Nur Zulaikha Binti Azhar, Noor Hasimah Ibrahim Teo, Raseeda binti Hamzah, Rosniza binti Roslan, and Ruhaila Maskat. A hybrid resnet—mobilenet deep learning model for smart bin waste classification. In 2024 5th International Conference on Artificial Intelligence and Data Sciences (AiDAS), pages 356–362. IEEE, 2024.
• Oriol Chac´on-Albero, Mario Campos-Mochol´ı, C´edric Marco-Detchart, Vicente Julian, Jaime Andr´es Rincon, and Vicent Botti. Ai for sustainable recycling: Efficient model optimization for waste classification systems. Sensors, 25(12):3807, 2025.
• Tio Dharmawan, Yudha Alif Auliya, Diah Ayu Retnani, Saiful Bukhori, Mohammad Zarkasi, and Imanuel Ataama. Optimizing densenet121 for waste classification using genetic algorithm-based downsampling and data augmentation. Mathematical Modelling of Engineering Problems, 12(5), 2025.
• Ricy Firnando, Muhammad Ali Buchari, Anna Dwi Marjusalinah, Rahmat Fadli Isnanto, et al. Performance analysis of mobilenetv2 based automatic waste classification using transfer learning. Jurnal Mandiri IT, 14(1):257–268, 2025.
• Akshat Gaurav, Brij Bhooshan Gupta, Varsha Arya, Razaz Waheeb Attar, Shavi Bansal, Ahmed Alhomoud, and Kwok Tai Chui. Smart waste classification in iot-enabled smart cities using vgg16 and cat swarm optimized random forest. Plos one, 20(2):e0316930, 2025.
• Hafizatul Hanin Hamzah, Wan Athirah Wan Endut, Azmina Mohd Ariffin, and Nur Atiqah Sia Abdullah. Waste management classification using convolutional neural network. In International Conference on Innovation & Entrepreneurship in Computing, Engineering & Science Education (InvENT 2024), pages 330–340. Atlantis Press, 2024.
• Teny Handhayani and Janson Hendryli. Leboh: An android mobile application for waste classification using tensorflow lite. In Proceedings of SAI Intelligent Systems Conference, pages 53–67. Springer, 2022.
• Teny Handhayani, Ageng Hadi Pawening, and Janson Hendryli. Leboh 2: An android application for solid waste detection. IAENG Int J Comput Sci, 50(4):1–9, 2023.
• Md Ibnul Hasan, Rabiei Mamat, and Mohammad Aizat Bin Basir. Advanced techniques for waste classification using a heterogeneous and hybrid ensemble of convolutional neural networks and parallel processing. Journal of Engineering and Science Research, 8(2):08–18, 2024.
• Md Mosarrof Hossen, Molla E Majid, Saad Bin Abul Kashem, Amith Khandakar, Mohammad Nashbat, Azad Ashraf, Mazhar Hasan-Zia, Ali K Ansaruddin Kunju, Saidul Kabir, and Muhammad EH Chowdhury. A reliable and robust deep learning model for effective recyclable waste classification. IEEE Access, 12:13809–13821, 2024.
• Kai Huang, Huan Lei, Zeyu Jiao, and Zhenyu Zhong. Recycling waste classification using vision transformer on portable device. Sustainability, 13(21):11572, 2021.
• Amir Mahmud Husein, Baren Baruna Harahap, Tio Fulalo Simatupang, Karunia Syukur Baeha, and Bintang Keitaro Sinambela. A hybrid vision transformer model for efficient waste classification. Jurnal Ilmu Komputer dan Informasi, 18(2):261–275, 2025.
• Md Minhazul Islam, SM Mahedy Hasan, Md Rakib Hossain, Md Palash Uddin, and Md Al Mamun. Towards sustainable solutions: Effective waste classification framework via enhanced deep convolutional neural networks. Plos one, 20(6):e0324294, 2025.
• Daniel Hogan Itam, Ekwueme Chimeme Martin, and Ibiba Taiwo Horsfall. Enhanced convolutional neural network methodology for solid waste classification utilizing data augmentation techniques. Waste Management Bulletin, 2(4):184– 193, 2024.
• Suman Kunwar. Mwaste: A deep learning approach to manage household waste. arXiv preprint arXiv:2304.14498, 2023.
• Suman Kunwar, Banji Raphael Owabumoye, and Abayomi Simeon Alade. Plastic waste classification using deep learning: insights from the wadaba dataset. arXiv preprint arXiv:2412.20232, 2024.
• Hiranya Jeet Malla, Milad Bazli, and Mehrdad Arashpour. Enhancing waste recognition with vision-language models: A prompt engineering approach for a scalable solution. Waste Management, 204:114939, 2025.
• Dishant Mewada, Eoin Martino Grua, Ciaran Eising, Patrick Denny, Pepijn Van de Ven, and Anthony Scanlan. Zero-shot learning for sustainable municipal waste classification. Recycling, 10(4):144, 2025.
• M. Mohamed. Garbage classification dataset. https://www.kaggle.com/datasets/mostafaabla/ garbage-classification, 2020. [Accessed 1 February 2026].
• F. Ozkefe. Trashnet dataset. https://www.kaggle.com/ datasets/feyzazkefe/trashnet, 2022. [Accessed 1 February 2026].
• Mujalin Polchan, Ampai Pukao, Tapana Cheunban, and Somnuek Sinthupuan. Waste management system using waste classification on mobile application. In 2023 Joint International Conference on Digital Arts, Media and Technology with ECTI Northern Section Conference on Electrical, Electronics, Computer and Telecommunications Engineering (ECTI DAMT & NCON), pages 229–233. IEEE, 2023.
• Wenxuan Qiu, Chenxin Xie, and Jingui Huang. An improved efficientnetv2 for garbage classification. In International Conference on Intelligent Computing, pages 79–90. Springer, 2025.
• Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4510–4520, 2018.
• Faizul Rakib Sayem, Md Sakib Bin Islam, Mansura Naznine, Mohammad Nashbat, Mazhar Hasan-Zia, Ali K Ansaruddin Kunju, Amith Khandakar, Azad Ashraf, Molla Ehsanul Majid, Saad Bin Abul Kashem, et al. Enhancing waste sorting and recycling efficiency: robust deep learning-based approach for classification and detection. Neural Computing and Applications, 37(6):4567–4583, 2025.
• C Vyas, K Rathod, K Makvana, A Nimavat, and K Kandoriya. Smart waste sorting using deep learning and image augmentation. African Journal of Biomedical Research, 27:5777–5785, 2024.
• Liying Yong, Le Ma, Dandan Sun, and Liping Du. Application of mobilenetv2 to waste classification. Plos one, 18(3):e0282336, 2023.
• Dashun Zheng, Rongsheng Wang, Yaofei Duan, Patrick Cheong-Iao Pang, and Tao Tan. Focus-rcnet: a lightweight recyclable waste classification algorithm based on focus and knowledge distillation. Visual Computing for Industry, Biomedicine, and Art, 6(1):19, 2023.
• Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592, 2023.