Classification of recycling waste using transfer learning

Öz

Nowadays, increasing industrialization and population have caused an increase in the amount of waste. Thus, waste management and recycling processes have gained great importance. Improved object waste detection systems play an important role in the correct classification of waste and in increasing the efficiency of recycling processes. In this study, transfer learning based YOLOv8 was used for object detection of recyclable waste. YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, YOLOv8x and RT-DETR models have been used to evaluate the success of the YOLO. In the experimental results, YOLOv8n was the most effective model is observed. In addition, real-time classification has been implemented by using waste images, video or webcam in the proposed model. 7963 different images were used for modeling. The trained model can detect frequently used objects including Battery, Bottle Cap, Cardboard, Garbage Bag, Glass Bottle, Plastic Bag, Plastic Bottle and Box. In the proposed method, recycling waste objects have been detected with an average accuracy rate (mAP) of 86.6%.

Anahtar Kelimeler

• Waste Management
• Transfer Learning
• Object Waste Detection
• Environmental Sustainability
• YOLO

Transfer öğrenme kullanılarak geri dönüşüm atıklarının sınıflandırılması

Öz

Günümüzde artan sanayileşme ve nüfus, atık sayısının artması sebep olmuştur. Dolayısıyla atık yönetimi ve geri dönüşüm süreçleri büyük bir önem kazanmaktadır. Geliştirilmiş nesne atık tespit sistemleri; atıkların doğru bir şekilde sınıflandırılmasını, geri dönüşüm süreçlerinin verimliliğini artırılmasında önemli rol oynamaktadır. Bu çalışmada, geri dönüşümlü atıkların nesne tespiti için transfer öğrenme temelli YOLOv8 kullanılmıştır. YOLO yönteminin başarısının değerlendirmek için YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, YOLOv8x ve RT-DETR modelleri kullanılmıştır. Gerçekleştirilen deneysel sonuçlarda en etkin olan modelin YOLOv8n olduğu gözlenmiştir. Ayrıca, önerilen modelde atık görüntüsü, video veya web kamerası kullanılarak gerçek zamanlı olarak sınıflandırılma gerçekleştirilmiştir. Modelleme için 7963 farklı görüntü kullanılmıştır. Eğitilmiş olan model ile geri dönüşümü hedeflenen sıklıkla kullanılan Batarya, Şişe Kapağı, Karton, Çöp Poşeti, Cam Şişe, Plastik Poşet, Plastik Şişe ve Teneke nesneleri tespit edilebilmektedir. Önerilen yöntem ile geri dönüşüm atık nesneleri %86.6 ortalama doğruluk oranı (mAP) ile tespit edilmiştir.

Anahtar Kelimeler

• Atık Yönetimi
• Transfer Öğrenme
• Nesne Atık Tespiti
• Çevresel Sürdürülebilirlik
• YOLO

Kaynakça

1. United Nations Environment Programme, Global Waste Management Outlook 2024: Beyond an age of waste – Turning rubbish into a resource. Nairobi, 2024. https://wedocs.unep.org/20.500.11822/44939.
2. U.S. Environmental Protection Agency, Global Greenhouse Gas Emissions Data, 2019. https://www.epa.gov/ghgemissions/global-greenhouse-gas-overview.
3. Department of Environment, Science and Innovation, Queensland, Economic Impacts, WetlandInfo website, 2021. https://wetlandinfo.des.qld.gov.au/wetlands/management/pressures/litter-illegal-dumping/effects-values/economic-impacts.html.
4. N. Ferronato ve V. Torretta, Waste mismanagement in developing countries: A review of global ıssues, International Journal of Environmental Research and Public Health, 16 (6), 1060, 2019. https://doi.org/10.3390/ijerph16061060.
5. J. Bobulski ve M. Kubanek, Deep learning for plastic waste classification system, Applied Computational Intelligence and Soft Computing, 2021, 6626948, 7 sayfa, 2021. https://doi.org/10.1155/2021/6626948.
6. N. Nnamoko, J. Barrowclough ve J. Procter, Solid waste ımage classification using deep convolutional neural network, Infrastructures, 7, 47, 2022. https://doi.org/10.3390/infrastructures7040047.
7. BBC Türkçe, Yapay zeka atıkları azaltmak ve geri dönüşümü artırmak için nasıl kullanılabilir?. https://www.bbc.com/turkce/articles/cn0k40m2ndjo.
8. Recycleye, https://recycleye.com/. Accessed 24 June 2024

9. A. Altınörs ve S. Çelik, YOLOv3 derin öğrenme algoritması ile İHA görüntülerinden çevresel atık tespiti, International Journal of Innovative Engineering Applications, 7 (1), 76-85, 2023. https://doi.org/10.46460/ijiea.1195428.
10. K. Andhy Panca Saputra ve K. Kusrini, Waste object detection and classification using deep learning algorithm: YOLOv4 and YOLOv4-tiny, Turkish Journal of Computer and Mathematics Education (TURCOMAT), 12 (14), 1666–1677, 2021. https://doi.org/10.17762/turcomat.v12i14.10504.
11. K. Erin, B. Bingöl ve B. Boru, YOLO – Based waste detection, Journal of Smart Systems Research, 3 (2), 120-127, 2022.
12. K. Demir ve O. Yaman, A YOLOv4-based method for underwater litter detection, International Informatics Congress (IIC2022) (Haz.), ss. 354-361, 2022.
13. X. Jiang, H. Hu, Y. Qin ve diğ., A real-time rural domestic garbage detection algorithm with an improved YOLOv5s network model, Sci Rep, 12, 16802, 2022. https://doi.org/10.1038/s41598-022-20983-1.
14. E. J. C. Flores, A YOLOv5s Model for classification of garbage, Southeast Asian Journal of Science and Technology, 8 (1), 9-18, 2023. https://sajst.org/online/index.php/sajst/article/view/285.
15. L. Chen ve J. Zhu, Water surface garbage detection based on lightweight YOLOv5, Sci Rep, 14 (1), 6133, 2024. https://doi.org/10.1038/s41598-024-55051-3.
16. S. Pan, N. Wang, Y. Lin ve J. Tang, Based on YOLOV8 ıntelligent trash can garbage classification detection algorithm, Mathematical Modeling and Algorithm Application, 2 (1), 28-32, 2024. https://doi.org/10.54097/67mska34.
17. I. Shukhratov, A. Pimenov, A. Stepanov, N. Mikhailova, A. Baldycheva ve A. Somov, Optical detection of plastic waste through computer vision, Intelligent Systems with Applications, 22, 200341, 2024. https://doi.org/10.1016/j.iswa.2024.200341.
18. M. Panmuang ve C. Rodmorn, Garbage detection using YOLO algorithm for urban management in Bangkok, WSEAS Transactions on Computer Research, 12, 236-243, 2024. https://doi.org/10.37394/232018.2024.12.23.
19. M. Zhu ve E. Kong, Multi-scale fusion uncrewed aerial vehicle detection based on RT-DETR, Electronics, 13 (8), 1489, 2024. https://doi.org/10.3390/electronics13081489.
20. Ultralytics, https://www.ultralytics.com/tr. Accessed 24 June 2024.
21. J. Terven, D. Córdova-Esparza ve J. Romero-González, A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS, Machine Learning and Knowledge Extraction, 5 (4), 1680–1716, Kasım 2023. https://doi.org/10.3390/make5040083.
22. Ultralytics. https://docs.ultralytics.com/tr/models/yolov8/. Accessed 24 June 2024.
23. Supported Tasks and Modes. https://docs.ultralytics.com/models/yolov8/#supported-tasks-and-modes. Accessed 24 June 2024.
24. Ultralytics. https://docs.ultralytics.com/tr/models/yolov8/#performance-metrics.
25. YOLOv8 Architecture: A Deep Dive into its Architecture. https://yolov8.org/yolov8-architecture/. Accessed 24 June 2024.
26. B. Yu, S. Li, W. Wang ve diğ., The DCN-BiFPN Object Detection Algorithm based on YOLOv8, 14 Mart 2024, PREPRINT (Versiyon 1), Research Square. https://doi.org/10.21203/rs.3.rs-4019974/v1.
27. R.-Y. Ju ve W. Cai, Fracture detection in pediatric wrist trauma X-ray images using YOLOv8 algorithm, Scientific Reports, 13, 2023. https://doi.org/10.1038/s41598-023-47460-7.
28. Y. Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y. Liu ve J. Chen, DETRs Beat YOLOs on Real-time Object Detection, arXiv. https://arxiv.org/abs/2304.08069.
29. Roboflow. https://roboflow.com/. Accessed 24 June 2024.
30. YOGI TRI A., Kardus Packing Dataset, Roboflow Universe, Kasım 2023. https://universe.roboflow.com/yogi-tri-a/kardus-packing/dataset/2. Accessed 24 June 2024.
31. packagedetection, package-detection Dataset, Roboflow Universe, Aralık 2023. https://universe.roboflow.com/packagedetection/package-detection-kt9ut/dataset/2.
32. YOGI TRI A., Box Carton Package Detection Dataset, Roboflow Universe, Kasım 2023. https://universe.roboflow.com/yogi-tri-a/box-carton-package-detection/dataset/6. Accessed 24 June 2024.
33. Chris Adam, Digital Capsx27;ule memories Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/chris-adam-b9cuh/digital-caps-ule-memories/browse?queryText=&pageSize=50&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
34. bottle cup detection, bottle cap color detection Dataset, Roboflow Universe, Ekim 2023. https://universe.roboflow.com/bottle-cup-detection-gnira/bottle-cap-color-detection/dataset/9. Accessed 24 June 2024.
35. AleksPilot, BottleI DEDataset, Roboflow Universe, Mayıs 2023. https://universe.roboflow.com/alekspilot-kvgn4/bottle_ide/dataset/4. Accessed 24 June 2024.
36. dogcat, glass bottle Dataset, Roboflow Universe, Şubat 2024. https://universe.roboflow.com/dogcat-nlhtn/glass-bottle-juhvb/dataset/1. Accessed 24 June 2024.
37. HTW Berlin University of Applied Sciences, plasticandglass Dataset, Roboflow Universe, Haziran 2023. https://universe.roboflow.com/htw-berlin-university-of-applied-sciences/plasticandglass/browse?queryText=&pageSize=200&startingIndex=200&browseQuery=true. Accessed 24 June 2024.
38. Beach Cleaning Robot, Glass detection Dataset, Roboflow Universe, Aralık 2023. https://universe.roboflow.com/beach-cleaning-robot/glass-detection-momkc/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
39. Arun Kumar, plastic Annotations Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/arun-kumar-wnuph/plastic-annotations/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
40. DenisWang, bottletest03Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/deniswang/bottle_test03/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
41. bottles, Bottles Dataset, Roboflow Universe, Ocak 2024. https://universe.roboflow.com/bottles-t9lzo/bottles-ramni/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
42. klasifikasi, botol 350 Dataset, Roboflow Universe, Nisan 2024. https://universe.roboflow.com/klasifikasi-ghgpr/botol-350/browse?queryText=&pageSize=200&startingIndex=200&browseQuery=true. Accessed 24 June 2024.
43. BEproject, BE Dataset, Roboflow Universe, Ocak 2024. https://universe.roboflow.com/beproject-ontnq/be-m6kyc/browse?queryText=&pageSize=200&startingIndex=200&browseQuery=true. Accessed 24 June 2024.
44. Project, Plastic object detection USING YOLOV8 Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/project-mwpp7/plastic-object-detection-using-yolov8/dataset/1. Accessed 24 June 2024.
45. MTU1, MTU1projectDataset, Roboflow Universe, Ocak 2024. https://universe.roboflow.com/mtu1-iu57z/mtu_1_project/dataset/1.
46. garbage bag, trashbag finder Dataset, Roboflow Universe, Nisan 2023. https://universe.roboflow.com/garbage-bag/trashbag-finder/dataset/2. Accessed 24 June 2024.
47. SnapCycleV2, Batteries Dataset, Roboflow Universe, Temmuz 2023. https://universe.roboflow.com/snapcyclev2/batteries-ephqt/browse?queryText=&pageSize=200&startingIndex=400&browseQuery=true. Accessed 24 June 2024.
48. ngoc411, battery Dataset, Roboflow Universe, Ocak 2024. https://universe.roboflow.com/ngoc411/battery-bgk7x/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true.
49. project-anaaa, battery Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/project-anaaa/battery-jrr08/browse?queryText=&pageSize=200&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
50. ELAKKIYA E., Plastic Litter Detection Dataset, Roboflow Universe, Ocak 2024. https://universe.roboflow.com/elakkiya-e-qijt0/plastic-litter-detection-yl09a/dataset/3. Accessed 24 June 2024.
51. mins workspace, ObjectDetectiondrinkDataset, Roboflow Universe, Ekim 2023. https://universe.roboflow.com/mins-workspace/objectdetection_drink/browse?queryText=&pageSize=200&startingIndex=200&browseQuery=true. Accessed 24 June 2024.
52. YoloV8, SolidWasteClassifier Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/yolov8-erigx/solidwasteclassifier/browse?queryText=&pageSize=200&startingIndex=800&browseQuery=true. Accessed 24 June 2024.
53. varyrecycle, can Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/varyrecycle/can-cwf4o/browse?queryText=class%3Acan&pageSize=200&startingIndex=200&browseQuery=true. Accessed 24 June 2024.
54. SnapCycle, SnapCycle Dataset, Roboflow Universe, Ağustos 2023. https://universe.roboflow.com/snapcycle/snapcycle/browse?queryText=class%3Ametal-can&pageSize=200&startingIndex=400&browseQuery=true. Accessed 24 June 2024.
55. Waste Detector, Cardboard-BottleCap-GlassBottle-PlasticBottle-PlasticBag-GarbageBag-Battery-Tin Dataset, Roboflow Universe, Haziran 2024. https://universe.roboflow.com/waste-detector-2t9ly/cardboard-bottlecap-glassbottle-plasticbottle-plasticbag-garbagebag-battery-tin/browse?queryText=class%3Anull&pageSize=50&startingIndex=0&browseQuery=true. Accessed 24 June 2024.
56. Karadağ, C., & Özdemir, D. (2023). Comparative analysis of deep learning methods for brain tumor detection. Artificial Intelligence Studies, 6(1), 1-13. https://doi.org/10.30855/AIS.2023.06.01.01
57. Google Colaboratory Pro. https://colab.research.google.com. Accessed 24 June 2024.