The use of camera traps is a nonintrusive monitoring method to obtain valuable information about the appearance and behavior of wild animals. However, each study generates thousands of pictures and extracting information remains mostly an expensive, time-consuming manual task. Nevertheless, image recognition and analyzing technologies combined with machine learning algorithms, particularly deep learning models, improve and speed up the analysis process. Therefore, we tested the usability of a pre-trained deep learning model available on the TensorFlow hub–FasterRCNN+InceptionResNet V2 network applied to images of ten different European wild mammal species such as wild boar (Sus scrofa), roe deer (Capreolus capreolus), or red fox (Vulpes vulpes) in color as well as black and white infrared images. We found that the detection rate of the correct region of interest (region of the animal) was 94%. The classification accuracy was 71% for the correct species’ name as mammals and 93% for the correct species or higher taxonomic ranks such as “carnivore” as order. In 7% of cases, the classification was incorrect as the wrong species’ name was classified. In this technical note, we have shown the potential of an existing and pre-trained image classification model for wildlife animal detection, classification, and analysis. A specific training of the model on European wild mammal species could further increase the detection and classification accuracy of the models. Analysis of camera trap images could thus become considerably faster, less expensive, and more efficient.
This is a preview of subscription content, access via your institution.
Buy single article
Instant access to the full article PDF.
Price excludes VAT (USA)
Tax calculation will be finalised during checkout.
Abadi M, Agarwal A, Barham P, Brevdo E, Chen Z, Citro C, Corrado G.S, Davis A, Dean J, Devin M (2015) TensorFlow: large-scale machine learning on heterogeneous systems. 2015. Available online: Tensorflow.org (accessed on 10.07.2019)
Banupriya N, Saranya S, Swaminathan R, Harikumar S, Palanisamy S (2020) Animal detection using deep learning algorithm. J Crit Rev 7(1):434–439
Beery S, Van Horn G, MacAodha O, Perona P (2019) The iWildCam 2018 Challenge Dataset. arXiv preprint arXiv:1904.05986. Available online: https://arxiv.org/pdf/1907.07617.pdf (accessed on 25.07.2019)
Bowkett AE, Rovero F, Marshall AR (2008) The use of camera-trap data to model habitat use by antelope species in the Udzungwa mountain forests, Tanzania. Afr J Ecol 46:479–487
Casaer J, Milotic T, Liefting Y, Desmet P, Jansen P (2019) Agouti: a platform for processing and archiving of camera trap images. Biodiversity Information Science and Standards. Available online: https://biss.pensoft.net/article/46690/ (accessed on 15.05.2020)
Figueroa K, Camarena-Ibarrola A, García J, Tejeda Villela H (2014) Fast automatic detection of wildlife in images from trap cameras. In: Bayro-Corrochano E, Hancock E (eds) Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications. CIARP 2014. Lecture Notes in Computer Science, vol 8827, pp 940–947
Google LLC (2019) Open Images Dataset V4, under CC BY 4.0 license. Available online: https://storage.googleapis.com/openimages/web/factsfigures_v4.html (accessed on 20.07.2019)
Google LLC - TensorFlow Hub (2019) Faster_rcnn/openimages_v4/inception_resnet_v2. Available online: https://tfhub.dev/google/faster_rcnn/openimages_v4/inception_resnet_v2/1 (accessed on 24.07.2019)
Google LLC Colaboratory (2019) Welcome to Colaboratory! Available online: https://colab.research.google.com/notebooks/welcome.ipynb (accessed on 20.07.2019)
He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770–778). Available online: https://arxiv.org/abs/1512.03385 (accessed on 24.07.2019)
Huang J, Rathod V, Sun C, Zhu M, Korattikara A, Fathi A, Murphy K (2017) Speed/accuracy trade-offs for modern convolutional object detectors. In proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7310-7311)
Hui J (2018) Object detection: speed and accuracy comparison (faster R-CNN, R-FCN, SSD, FPN, RetinaNet and YOLOv3). Medium. Available online: https://medium.com/@jonathan_hui/object-detection-speed-and-accuracy-comparison-faster-r-cnn-r-fcn-ssd-and-yolo-5425656ae359 (accessed on 25.07.2019)
Hunter JD (2007) Matplotlib: A 2D graphics environment. Comput Sci Eng 9:90–95 Available online: https://ieeexplore.ieee.org/document/4160265 (accessed on 24.05.2019)
Inik Ö, Turan B (2018) Classification of animals with different deep learning models. J New Results Sci 7(1):9–16
Kluyver T, Ragan-Kelley B, Pérez F, Granger BE, Bussonnier M, Frederic J, Kelly K, Hamrick J, Grout J, Corlay S, Jupyter Development Team (2016) Jupyter notebooks—a publishing format for reproducible computational workflows. ELPUB 2016:87–90 Available online: http://ebooks.iospress.nl/publication/42900 (accessed on 24.04.2019)
LILA BC (2019) Labeled Information Library of Alexandria: Biology and Conservation. Datasets. Available online: http://lila.science/datasets (accessed on 25.07.2019)
Lundh F, Clark A, Contributors (2016) Pillow - Python Imaging Library. Available online: https://pillow.readthedocs.io/en/3.1.x/index.html (accessed on 24.07.2019)
Miao Z, Gaynor KM, Wang J, Liu Z, Muellerklein O, Norouzzadeh MS, Getz WM (2019) Insights and approaches using deep learning to classify wildlife. Sci Rep 9(1):1–9
Nationalparkverwaltung Hainich, FFK Gotha (2019) Schwarzwildforschung im Hainich. Available online: https://www.schwarzwild-hainich.de (accessed on 24.07.2019)
Newey S, Davidson P, Nazir S, Fairhurst G, Verdicchio F, Irvine RJ, van der Wal R (2015) Limitations of recreational camera traps for wildlife management and conservation research: a practitioner’s perspective. Ambio 44(Suppl 4):624–635. https://doi.org/10.1007/s13280-015-0713-1
Norouzzadeh MS, Nguyen A, Kosmala M, Swanson A, Palmer MS, Packer C, Clune J (2018) Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. Proc Natl Acad Sci 115(25):E5716–E5725
Oliphant TE (2006) A Guide to NumPy, vol 1. Trelgol Publishing, Spanish Fork, p 85
Python Software Foundation (2019a) The Python Standard Library - tempfile — Generate temporary files and directories. Available online: https://docs.python.org/3/library/tempfile.html (accessed on 24.04.2019)
Python Software Foundation (2019b) The Python Standard Library - urllib — Open arbitrary resources by URL. Available online: https://docs.python.org/2/library/urllib.html (accessed on 24.05.2019)
Python Software Foundation (2019c) The Python Standard Library - io — Core tools for working with streams. Available online: https://docs.python.org/3/library/io.html (accessed on 24.05.2019)
Python Software Foundation (2019d) The Python Standard Library - time — Time access and conversions. Available online: https://docs.python.org/3/library/time.html (accessed on 24.05.2019)
Ren S, He K, Girshick R, Sun J (2015) Faster r-cnn: towards real-time object detection with region proposal networks. In Advances in neural information processing systems (pp. 91–99). Available online: https://arxiv.org/pdf/1506.01497.pdf (accessed on 24.05.2019)
Silveira L, Jacomo AT, Diniz-Filho JAF (2003) Camera trap, line transect census and track surveys: a comparative evaluation. Biol Conserv 114:351–355
Weingarth K, Bufka L, Daniszova K, Knauer F, Sustr P, Heurich M (2011) Nationalpark Bayerischer Wald (2011) Grenzüberschreitendes Fotofallenmonitoring - wie zählt man Luchse?. Available online: https://www.researchgate.net/publication/236963693_Grenzuberschreitendes_Fotofallenmonitoring_Wie_zahlt_man_Luchse (accessed on 22.02.2020)
Yu X, Wang J, Kays R, Jansen PA, Wang T, Huang T (2013) Automated identification of animal species in camera trap images. EURASIP J Image Video Process 2013:52–61 http://jivp.eurasipjournals.com/content/2013/1/52. Accessed 20 June 2019
We would like to thank the developer of the open-source Tensorflow models and the team of the Open Images Dataset V4 as well as the Hainich National Park for providing camera trap images.
This research was supported by the University of Applied Sciences Erfurt (FHE).
Conflict of interest
The authors declare that they have no conflict of interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Carl, C., Schönfeld, F., Profft, I. et al. Automated detection of European wild mammal species in camera trap images with an existing and pre-trained computer vision model. Eur J Wildl Res 66, 62 (2020). https://doi.org/10.1007/s10344-020-01404-y
- Computer vision
- Image analysis
- Camera trap
- Pre-trained model
- Wild mammal species