Animal Systematics, Evolution and Diversity

  • 제40권4호
  • /
  • Pages.266-275
  • /
  • 2024
  • /
  • 2234-6953(pISSN)
  • /
  • 2234-8190(eISSN)
한국동물분류학회 (The Korean Society of Systematic Zoology)
권호 표지
DOI QR코드

DOI QR Code

Application of Deep Learning-Based Object Detection Models to Classify Images of Cacatua Parrot Species

  • Jung-Il Kim (Department of Biotechnology, Sangmyung University) ;
  • Jong-Won Baek (Department of Biotechnology, Sangmyung University) ;
  • Chang-Bae Kim (Department of Biotechnology, Sangmyung University)
  • 투고 : 2024.06.12
  • 심사 : 2024.09.25
  • 발행 : 2024.10.31
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

Parrots, especially the Cacatua species, are a particular focus for trade because of their mimicry, plumage, and intelligence. Indeed, Cacatua species are imported most into Korea. To manage trade in wildlife, it is essential to identify the traded species. This is conventionally achieved by morphological identification by experts, but the increasing volume of trade is overwhelming them. Identification of parrots, particularly Cacatua species, is difficult due to their similar features, leading to frequent misidentification. There is thus a need for tools to assist experts in accurately identifying Cacatua species in situ. Deep learning-based object detection models, such as the You Only Look Once (YOLO) series, have been successfully employed to classify wildlife and can help experts by reducing their workloads. Among these models, YOLO versions 5 and 8 have been widely applied for wildlife classification. The later model normally performs better, but selecting and designing a suitable model remains crucial for custom datasets, such as wildlife. Here, YOLO versions 5 and 8 were employed to classify 13 Cacatua species in the image data. Images of these species were collected from eBird, iNaturalist, and Google. The dataset was divided, with 80% used for training and validation and 20% for evaluating model performance. Model performance was measured by mean average precision, with YOLOv5 achieving 0.889 and YOLOv8 achieving 0.919. YOLOv8 was thus better than YOLOv5 at detecting and classifying Cacatua species in the examined images. The model developed here could significantly support the management of the global trade in Cacatua species.

키워드

과제정보

This work was supported by a grant from the National Institute of Biological Resources(NIBR), funded by the Ministry of Environment (MOE) of the Republic of Korea (NIBRE 202411). We used the images from the Macaulay Library at the Cornell Lab of Ornithology during model training, and we thank the thousands of eBird participants and organizations for their contributions.

참고문헌

  1. Ahmad I, Yang Y, Yue Y, Ye C, Hassan M, Cheng X, Wu Y, Zhang Y, 2022. Deep learning based detector YOLOv5 for identifying insect pests. Applied Sciences, 12:10167. https://doi.org/10.3390/app121910167
  2. Aloysius SLM, Yong DL, Lee JG, Jain A, 2020. Flying into extinction: understanding the role of Singapore's international parrot trade in growing domestic demand. Bird Conservation International, 30:139-155. https://doi.org/10.1017/S0959270919000182
  3. Alzubaidi L, Bai J, Al-Sabaawi A, Santamaria J, Albahri AS, Al-dabbagh BSN, Fadhel MA, Manoufali M, Zhang J, Al-Timemy AH, Duan Y, Abdullah A, Farhan L, Lu Y, Gupta A, Albu F, Abbosh A, Gu Y, 2023. A survey on deep learning tools dealing with data scarcity: definitions, challenges, solutions, tips, and applications. Journal of Big Data, 10:46. https://doi.org/10.1186/s40537-023-00727-2
  4. Andersson AA, Gibson L, Baker DM, Cybulski JD, Wang S, Leung B, Chu LM, Dingle C, 2021. Stable isotope analysis as a tool to detect illegal trade in critically endangered cockatoos. Animal Conservation, 24:1021-1031. https://doi.org/10.1111/acv.12705
  5. Austrheim G, Speed JDM, Martinsen V, Mulder J, Mysterud A, 2014. Experimental effects of herbivore density on aboveground plant biomass in an alpine grassland ecosystem. Arctic, Antarctic, and Alpine Research, 46:535-541. https://doi.org/10.1657/1938-4246-46.3.535
  6. Baek JW, Kim JI, Kim CB, 2023. Deep learning-based image classification of turtles imported into Korea. Scientific Reports, 13:21677. https://doi.org/10.1038/s41598-023-49022-3
  7. Berg ML, Bennett ATD, 2010. The evolution of plumage colouration in parrots: a review. Emu-Austral Ornithology, 110:10-20. https://doi.org/10.1071/MU09076
  8. Bjerge K, Alison J, Dyrmann M, Frigaard CE, Mann HMR, Hoye TT, 2023. Accurate detection and identification of insects from camera trap images with deep learning. PLOS Sustainability and Transformation, 2:e0000051. https://doi.org/10.1371/journal.pstr.0000051
  9. Bradbury JW, Balsby TJS, 2016. The functions of vocal learning in parrots. Behavioral Ecology and Sociobiology, 70:293-312. https://doi.org/10.1007/s00265-016-2068-4
  10. Brandis KJ, Meagher P, Schoppe S, Zawada K, Widmann I, Widmann P, Dolorosa RG, Francis R, 2023. Determining the provenance of traded wildlife in the Philippines. Animals, 13:2165. https://doi.org/10.3390/ani13132165
  11. Bush ER, Baker SE, Macdonald DW, 2014. Global trade in exotic pets 2006-2012. Conservation Biology, 28:663-676. https://doi.org/10.1111/cobi.12240
  12. Buslaev A, Iglovikov VI, Khvedchenya E, Parinov A, Druzhinin M, Kalinin AA, 2020. Albumentations: fast and flexible image augmentations. Information, 11:125. https://doi.org/10.3390/info11020125
  13. Cardoso AS, Bryukhova S, Renna F, Reino L, Xu C, Xiao Z, Correia R, Di Minin E, Ribeiro J, Vaz AS, 2023. Detecting wildlife trafficking in images from online platforms: a test case using deep learning with pangolin images. Biological Conservation, 279:109905. https://doi.org/10.1016/j.biocon.2023.109905
  14. Chan DTC, Poon ESK, Wong ATC, Sin SYW, 2021. Global trade in parrots: influential factors of trade and implications for conservation. Global Ecology and Conservation, 30:e01784. https://doi.org/10.1016/j.gecco.2021.e01784
  15. Chan WH, Fung BSB, Tsang DHK, Lo IMC, 2023. A freshwater algae classification system based on machine learning with StyleGAN2-ADA augmentation for limited and imbalanced datasets. Water Research, 243:120409. https://doi.org/10.1016/j.watres.2023.120409
  16. Chowdhury S, Fuller RA, Ahmed S, Alam S, Callaghan CT, Das P, Correia RA, Di Marco M, Di Minin E, Jaric I, Labi MM, Ladle RJ, Rokonuzzaman M, Roll U, Sbragaglia V, Siddika A, Bonn A, 2023. Using social media records to inform conservation planning. Conservation Biology, 38:e14161.
  17. Collar NJ, Marsden SJ, 2014. The subspecies of Yellow-crested Cockatoo Cacatua sulphurea. Forktail, 30:23-27.
  18. Cussen VA, 2017. Psittacine cognition: individual differences and sources of variation. Behavioural Processes, 134:103-109. https://doi.org/10.1016/j.beproc.2016.11.008
  19. Dang F, Chen D, Lu Y, Li Z, 2023. YOLOWeeds: a novel benchmark of YOLO object detectors for multi-class weed detection in cotton production systems. Computers and Electronics in Agriculture, 205:107655. https://doi.org/10.1016/j.compag.2023.107655
  20. Del Hoyo J, 2020. All the birds of the world. Lynx Edicions, Barcelona, pp. 364-366.
  21. Esipova O, Love E, Noakes A, Schatz A, Swartz K, Vallianos C, 2021. Wildlife trafficking detection tools: best practices and application to the illegal rhino horn trade. Conservation International, Arlington, VA, USA, pp. 30-33.
  22. Forshaw JM, 2010. Parrots of the world. Princeton University Press, Princeton, NJ, pp. 1-336.
  23. Furnell S, 2019. Strengthening CITES processes for reviewing trade in captive-bred specimens and preventing mis-declaration and laundering: a review of trade in Southeast Asian parrot species [Internet]. BirdLife International, Cambridge, Accessed 12 Mar 2024, .
  24. Huerta-Ramos G, Lustrik R, 2021. Inat_Images: v.1.1 [Internet]. Zenodo, Accessed 12 Mar 2024, .
  25. Hughes AC, 2021. Wildlife trade. Current Biology, 31:R1218-R1224. https://doi.org/10.1016/j.cub.2021.08.056
  26. Jang W, Lee EC, 2021. Multi-class parrot image classification including subspecies with similar appearance. Biology, 10:1140. https://doi.org/10.3390/biology10111140
  27. Jocher G, Chaurasia A, Qiu J, 2023. Ultralytics YOLOv8 [Internet]. Github, San Francisco, CA, Accessed 16 Mar 2024, .
  28. Jocher G, Chaurasia A, Stoken A, Borovec J, Kwon Y, Michael K, Fang J, Wong C, Yifu Z, Montes D, 2022. ultralytics/yolov5: v6. 2-yolov5 classification models, apple m1, reproducibility, clearml and deci. ai integrations[Internet]. Github, San Francisco, CA, Accessed 16 Mar 2024, .
  29. Kim JI, Baek JW, Kim CB, 2022. Image classification of amazon parrots by deep learning: a potentially useful tool for wildlife conservation. Biology, 11:1303. https://doi.org/10.3390/biology11091303
  30. Kim YG, 2004. AutoCrawler [Internet]. Github, San Francisco, CA, Accessed 2 Mar 2024, .
  31. Kulkarni R, Di Minin E, 2023. Towards automatic detection of wildlife trade using machine vision models. Biological Conservation, 279:109924. https://doi.org/10.1016/j.biocon.2023.109924
  32. Mahendiran M, Parthiban M, Azeez PA, Nagarajan R, 2018. In situ measurements of animal morphological features: a non-invasive method. Methods in Ecology and Evolution, 9:613-623. https://doi.org/10.1111/2041-210X.12898
  33. McClure EC, Sievers M, Brown CJ, Buelow CA, Ditria EM, Hayes MA, Pearson RM, Tulloch VJ, Unsworth RKF, Connolly RM, 2020. Artificial intelligence meets citizen science to supercharge ecological monitoring. Patterns, 1:100109. https://doi.org/10.1016/j.patter.2020.100109
  34. Paramita C, Supriyanto C, Putra KR, 2024. Comparative analysis of YOLOv5 and YOLOv8 cigarette detection in social media content. Scientific Journal of Informatics, 11:341-352. https://doi.org/10.15294/sji.v11i2.2808
  35. Parr M, Juniper T, 2010. Parrots: a guide to parrots of the world. Bloomsbury Publishing, London, pp. 1-436.
  36. Roy AM, Bhaduri J, Kumar T, Raj K, 2023. WilDect-YOLO: an efficient and robust computer vision-based accurate object localization model for automated endangered wildlife detection. Ecological Informatics, 75:101919. https://doi.org/10.1016/j.ecoinf.2022.101919
  37. Sary IP, Andromeda S, Armin EU, 2023. Performance comparison of YOLOv5 and YOLOv8 architectures in human detection using aerial images. Ultima Computing: Jurnal Sistem Komputer, 15:8-13. https://doi.org/10.31937/sk.v15i1.3204
  38. Scheffers BR, Oliveira BF, Lamb I, Edwards DP, 2019. Global wildlife trade across the tree of life. Science, 366:71-76. https://doi.org/10.1126/science.aav5327
  39. Seomun H, Kim HJ, 2016. The guideline for import and export review of CITES species. National Institute of Biological Resources, Incheon, pp. 76-132.
  40. Sheng B, Wu L, Zhang N, 2023. A maturity detection method for Hemerocallis citrina Baroni based on lightweight and attention mechanism. Applied Sciences, 13:12043. https://doi.org/10.3390/app132112043
  41. Shetty AD, Ashwath S, 2023. Animal detection and classification in image & video frames using YOLOv5 and YOLOv8. In: 2023 7th International Conference on Electronics, Communication and Aerospace Technology (ICECA). Institute of Electrical and Electronics Engineers, New York, pp. 677-683.
  42. Spiesman BJ, Gratton C, Hatfield RG, Hsu WH, Jepsen S, McCornack B, Patel K, Wang G, 2021. Assessing the potential for deep learning and computer vision to identify bumble bee species from images. Scientific Reports, 11:7580. https://doi.org/10.1038/s41598-021-87210-1
  43. Tannous M, Stefanini C, Romano D, 2023. A deep-learning-based detection approach for the identification of insect species of economic importance. Insects, 14:148. https://doi.org/10.3390/insects14020148
  44. Tian Z, Shen C, Chen H, He T, 2019. FCOS: fully convolutional one-stage object detection. Preprint at arXiv: https://doi.org/10.48550/arXiv.1904.01355
  45. Trail PW, 2021. Morphological analysis: a powerful tool in wildlife forensic biology. Forensic Science International: Animals and Environments, 1:100025. https://doi.org/10.1016/j.fsiae.2021.100025
  46. Tzutalin D, 2015. LabelImg. version 1.8.1 [Internet]. GitHub, San Francisco, CA, Accessed 9 Mar 2024, .
  47. Vo HT, Thien NN, Mui KC, 2023. Bird detection and species classification: using YOLOv5 and deep transfer learning models. International Journal of Advanced Computer Science and Applications, 14:939-947. https://doi.org/10.14569/IJACSA.2023.01407102
  48. Wang CY, Yeh IH, Liao HYM, 2024. YOLOv9: learning what you want to learn using programmable gradient information. Preprint at arXiv: https://doi.org/10.48550/arXiv.2402.13616
  49. Wang G, Chen Y, An P, Hong H, Hu J, Huang T, 2023. UAV-YOLOv8: a small-object-detection model based on improved YOLOv8 for UAV aerial photography scenarios. Sensors, 23:7190. https://doi.org/10.3390/s23167190
  50. Yan B, Fan P, Lei X, Liu Z, Yang F, 2021. A real-time apple targets detection method for picking robot based on improved YOLOv5. Remote Sensing, 13:1619. https://doi.org/10.3390/rs13091619
  51. Zhang Q, Yi X, Guo J, Tang Y, Feng T, Liu R, 2023. A few-shot rare wildlife image classification method based on style migration data augmentation. Ecological Informatics, 77:102237. https://doi.org/10.1016/j.ecoinf.2023.102237
  52. Zhong Y, Wang J, Peng J, Zhang L, 2020. Anchor box optimization for object detection. Preprint at arXiv: https://doi.org/10.48550/arXiv.1812.00469