A Model for Identifying the Forms of Discarded Products and Assessing Their Redesign Potential Based on Deep Learning
DOI:
https://doi.org/10.67541/istaer2624Keywords:
Redesign of discarded products; Shape recognition; Deep learning; Potential assessment; Graph neural networkAbstract
The high-value redesign of discarded products relies on an accurate assessment of their remaining structure and redesign potential. However, existing visual recognition methods struggle to handle product damage, deformation, and occlusion, and lack quantitative assessment methods for redesign value. To address these issues, this paper proposes an end-to-end deep learning joint model that simultaneously performs shape recognition and redesign potential assessment for discarded products. In the shape recognition branch, a multi-branch feature extraction structure and a deformation-robust attention mechanism are designed to capture features of distorted and missing areas. Additionally, a shape structure preservation loss and an edge contrast loss are introduced to enhance boundary segmentation ability. In the potential assessment branch, a three-element index system comprising structural reusability, material integrity, and shape modification freedom is constructed. The spatial topological relationships among components are encoded using graph neural networks, and a shape–potential joint embedding space is established through cross-modal fusion. Continuous potential scores and discrete levels are output in parallel. The model achieves dual-task joint optimization using gradient coordination and an uncertainty weighting strategy. Experiments on our self-built dataset of 12,847 images of discarded electronic products show that the proposed method achieves 66.2% mIoU and 61.4% mAP in the shape recognition task, which are 3.8 and 4.7 percentage points higher than those of Mask2Former, respectively. The local mIoU on damaged areas is improved by 10.7 percentage points to 57.6%. The Spearman rank correlation coefficient between the predicted potential and expert scoring is 0.812, and the mean absolute error is reduced to 0.106. This method provides a quantifiable intelligent decision-making tool for the re-design of discarded products.
References
[1] Dumée, L. F. (2022). Circular materials and circular design—review on challenges towards sustainable manufacturing and recycling. Circular economy and sustainability, 2(1), 9-23. https://doi.org/10.1007/s43615-021-00085-2
[2] Mesa, J. A., & González-Quiroga, A. (2023). Development of a diagnostic tool for product circularity: a redesign approach. Research in Engineering Design, 34(4), 401-420. https://doi.org/10.1007/s00163-023-00415-5
[3] Triguero, A., Moreno‐Mondéjar, L., & Sáez‐Martínez, F. J. (2023). Circular economy and firm performance: The influence of product life cycle analysis, upcycling, and redesign. Sustainable Development, 31(4), 2318-2331. https://doi.org/10.1002/sd.2509
[4] Boy, G. A., Masson, D., Durnerin, É., & Morel, C. (2024). PRODEC for human systems integration of increasingly autonomous systems. Systems engineering, 27(5), 805-826. https://doi.org/10.1002/sys.21751
[5] Yan, Y., Tong, L., Song, K., Tian, H., Man, Y., & Yang, W. (2023). SISG-Net: Simultaneous instance segmentation and grasp detection for robot grasp in clutter. Advanced Engineering Informatics, 58, 102189. https://doi.org/10.1016/j.aei.2023.102189
[6] Tian, H., Song, K., Tong, L., Man, Y., & Yan, Y. (2023). Robot unknown objects instance segmentation based on collaborative weight assignment RGB–depth fusion strategy. IEEE/ASME Transactions on Mechatronics, 29(3), 2032-2043.
https://doi.org/10.1109/tmech.2023.3327865
[7] Muhammad, K., Hussain, T., Ullah, H., Del Ser, J., Rezaei, M., Kumar, N., ... & De Albuquerque, V. H. C. (2022). Vision-based semantic segmentation in scene understanding for autonomous driving: Recent achievements, challenges, and outlooks. IEEE Transactions on Intelligent Transportation Systems, 23(12), 22694-22715. https://doi.org/10.1109/tits.2022.3207665
[8] Dumée, L. F. (2022). Circular materials and circular design—review on challenges towards sustainable manufacturing and recycling. Circular economy and sustainability, 2(1), 9-23. https://doi.org/10.1007/s43615-021-00085-2
[9] Strzelecka, C. (2024). Critical data studies meets discard studies: Waste data reflectivity in digital urban waste tracking system. Convergence, 30(6), 2109-2130. https://doi.org/10.2139/ssrn.4981045
[10] Zhang, X., Song, Y., Song, T., Yang, D., Ye, Y., Zhou, J., & Zhang, L. (2024). LDConv: Linear deformable convolution for improving convolutional neural networks. Image and Vision Computing, 149, 105190. https://doi.org/10.48550/arXiv.2311.11587
[11] Gao, F., Cai, C., Tang, W., Tian, Y., & Huang, K. (2024). RA2DC-Net: A residual augment-convolutions and adaptive deformable convolution for points-based anchor-free orientation detection network in remote sensing images. Expert Systems with Applications, 238, 122299. https://doi.org/10.1016/j.eswa.2023.122299
[12] Ravichandran, B., Boissie, K., & Madhavan, P. (2025). A multidimensional methodological approach to disassembly index calculation: Enhancing product dismountability and sustainability. Sustainable Production and Consumption, 56, 80-93. https://doi.org/10.1016/j.spc.2025.03.016
[13] Glas, M., Kersten, L., Mulders, T., Pernul, G., & Allodi, L. (2026, April). T1GER: An Instructional Re-Design of a Cyber Range Exercise in a Commercial Security Operations Center. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (pp. 1-22). https://doi.org/10.1145/3772318.3791226
[14] Orso, V., Ziviani, R., Bacchiega, G., Bondani, G., Spagnolli, A., & Gamberini, L. (2022). Employee-centric innovation: Integrating participatory design and video-analysis to foster the transition to Industry 5.0. Computers & Industrial Engineering, 173, 108661. https://doi.org/10.1016/j.cie.2022.108661
[15] Hercog, D., Bencak, P., Vincetič, U., & Lerher, T. (2022). Product assembly assistance system based on pick-to-light and computer vision technology. Sensors, 22(24), 9769.
https://doi.org/10.3390/s22249769
[16] Wang, Y., Peng, X., Huang, W., Ye, X., & Jiang, M. (2023). Self‐supervised non‐rigid structure from motion with improved training of Wasserstein GANs. IET Computer Vision, 17(4), 404-414. https://doi.org/10.1049/cvi2.12175
[17] Ghoul, A., Pan, J., Lingg, A., Kübler, J., Krumm, P., Hammernik, K., ... & Küstner, T. (2024). Attention-aware non-rigid image registration for accelerated MR imaging. IEEE transactions on medical imaging, 43(8), 3013-3026. https://doi.org/10.1109/tmi.2024.3385024
[18] Zhu, D., Wang, W., Xue, X., Xie, H., Cheng, G., & Wang, F. L. (2024). Structure-preserving image smoothing via contrastive learning. The Visual Computer, 40(8), 5139-5153. https://doi.org/10.1007/s00371-023-02897-9
[19] Pan, S., Pu, Y., Zhao, Z., Yang, Q., Gu, J., & Xu, D. (2025). StrCCL: Structure-Aware Contrastive Consistency Loss for Artistic Style Transfer. Expert Systems with Applications, 129901. https://doi.org/10.1016/j.eswa.2025.129901
[20] Cao, Z., Huang, Y., Zhou, Y., Zhou, P., & He, Y. (2025). GlogSemiFNet: a semi-supervised contrastive learning framework for high-resolution remote sensing farmland edge detection based on frequency domain filters. Geo-spatial Information Science, 1-25. https://doi.org/10.1080/10095020.2025.2589636
[21] Ruschhaupt, S., Troles, J., & Schmid, U. M. (2025). Comparing Mask R-CNN and Mask2Former architectures for individual tree crown delineation. https://doi.org/10.18420/giljt2025_13
[22] Alshawabkeh, S., Wu, L., Dong, D., Cheng, Y., & Li, L. (2025). A hybrid approach for pavement crack detection using mask R-CNN and vision transformer model. Computers, Materials, & Continua, 82(1), 561. https://doi.org/10.32604/cmc.2024.057213
[23] Shanmugasundaram, S., & Palaniappan, N. (2024). Improvement of Bounding Box and Instance Segmentation Accuracy Using ResNet-152 FPN with Modulated Deformable ConvNets v2 Backbone-based Mask Scoring R-CNN. International Journal of Image and Graphics, 24(06), 2450054. https://doi.org/10.1142/s0219467824500542
[24] Mesa, J. A., & González-Quiroga, A. (2023). Development of a diagnostic tool for product circularity: a redesign approach. Research in Engineering Design, 34(4), 401-420. https://doi.org/10.1007/s00163-023-00415-5
[25] Fanicchia, F., & Karlsdottir, S. N. (2023). Research and development on coatings and paints for geothermal environments: A review. Advanced Materials Technologies, 8(18), 2202031. https://doi.org/10.1002/admt.202202031
[26] Ferrero, V., DuPont, B., Hassani, K., & Grandi, D. (2022). Classifying component function in product assemblies with graph neural networks. Journal of Mechanical Design, 144(2), 021406.
https://doi.org/10.1115/1.4052720
[27] Jian, C., Lu, Y., Lin, M., & Zhang, M. (2023). A novel graph neural networks approach for 3D product model retrieval. International Journal of Computer Integrated Manufacturing, 36(3), 381-392. https://doi.org/10.1080/0951192x.2022.2104460
[28] Murat, A., & Ali, S. (2024). Low-resource POS tagging with deep affix representation and multi-head attention. IEEE Access, 12, 66495-66504. https://doi.org/10.1109/access.2024.3395454
[29] Sivaramakrishnan, J., Lee, G., Parlevliet, D., & Wong, K. W. (2022). Margin-based approach for outlier detection of industrial design data using a modified general regression neural network. AI EDAM, 36, e7. https://doi.org/10.1017/S0890060421000329
[30] Jadon, A., Patil, A., & Jadon, S. (2024, January). A comprehensive survey of regression-based loss functions for time series forecasting. In International Conference on Data Management, Analytics & Innovation (pp. 117-147). Singapore: Springer Nature Singapore. https://doi.org/10.1007/978-981-97-3245-6_9
[31] Shi, Z., Yao, W., Zhao, Y., Ai, X., Wen, J., & Cheng, S. (2024). Two-stage weakly supervised learning to mitigate label noise for intelligent identification of power system dominant instability mode. Applied Energy, 359, 122659. https://doi.org/10.1016/j.apenergy.2024.122659
[32] Ye, Z., Wang, X., & Cai, Z. (2025). Performance Evaluation and Optimization of Multi-stage Manufacturing Systems: A Review. Reliability Analysis and Maintenance Optimization of Complex Systems: Essays in Honor of Professor Won Young Yun on his 65th Birthday, 211-243. https://doi.org/10.1007/978-3-031-70288-4_13
[33] Nath, D., Ankit, Neog, D. R., & Gautam, S. S. (2024). Application of Machine Learning and Deep Learning in Finite Element Analysis: A Comprehensive Review: D. Nath et al. Archives of computational methods in engineering, 31(5), 2945-2984.
https://doi.org/10.1007/s11831-024-10063-0
[34] Aliferis, C., & Simon, G. (2024). Overfitting, underfitting and general model overconfidence and under-performance pitfalls and best practices in machine learning and AI. Artificial intelligence and machine learning in health care and medical sciences: Best practices and pitfalls, 477-524. https://doi.org/10.1007/978-3-031-39355-6_10
[35] Pinto, A. H., Lounis, C., Causse, M., & Chanel, C. P. (2025). Online Metrics to Enhance Human-Artificial Agent Collaboration Efficiency: A Narrative Literature Review. International Journal of Human–Computer Interaction, 1-24. https://doi.org/10.1080/10447318.2025.2575895
Published
Data Availability Statement
The data that support the findings of this study are available upon request from the corresponding authors, M.Y.
Issue
Section
License
Copyright (c) 2026 Mingxiang Yang (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
X (FORMERLY TWITTER)