Skip to main content
×
×
Home

‘Tailception’: using neural networks for assessing tail lesions on pictures of pig carcasses

  • J. Brünger (a1), S. Dippel (a2), R. Koch (a1) and C. Veit (a2)
Abstract

Tail lesions caused by tail biting are a widespread welfare issue in pig husbandry. Determining their prevalence currently involves labour intensive, subjective scoring methods. Increased societal interest in tail lesions requires fast, reliable and cheap systems for assessing tail status. In the present study, we aimed to test the reliability of neural networks for assessing tail pictures from carcasses against trained human observers. Three trained observers scored tail lesions from automatically recorded pictures of 13 124 pigs. Nearly all pigs had been tail docked. Tail lesions were classified using a 4-point score (0=no lesion, to 3=severe lesion). In addition, total tail loss was recorded. Agreement between observers was tested prior and during the assessment in a total of seven inter-observer tests with 80 pictures each. We calculated agreement between observer pairs as exact agreement (%) and prevalence-adjusted bias-adjusted κ (PABAK; value 1=optimal agreement). Out of the 13 124 scored pictures, we used 80% for training and 20% for validating our neural networks. As the position of the tail in the pictures varied (high, low, left, right), we first trained a part detection network to find the tail in the picture and select a rectangular part of the picture which includes the tail. We then trained a classification network to categorise tail lesion severity using pictures scored by human observers whereby the classification network only analysed the selected picture parts. Median exact agreement between the three observers was 80% for tail lesions and 94% for tail loss. Median PABAK for tail lesions and loss were 0.75 and 0.87, respectively. The agreement between classification by the neural network and human observers was 74% for tail lesions and 95% for tail loss. In other words, the agreement between the networks and human observers were very similar to the agreement between human observers. The main reason for disagreement between observers and thereby higher variation in network training material were picture quality issues. Therefore, we expect even better results for neural network application to tail lesions if training is based on high quality pictures. Very reliable and repeatable tail lesion assessment from pictures would allow automated tail classification of all pigs slaughtered, which is something that some animal welfare labels would like to do.

Copyright
Corresponding author
Footnotes
Hide All
a

These three authors contributed equally to the research and article preparation.

b

Present address: Department of Food Safety and Infection Biology, Faculty of Veterinary Medicine, Norwegian University of Life Sciences, Ullevålsveien 72, 0454 Oslo, Norway.

Footnotes
References
Hide All
Blömke, L and Kemper, N 2017. Automated assessment of animal welfare indicators in pigs at slaughter. In Proceedings of the 12th International Symposium on the Epidemiology and Control of Biological, Chemical and Physical Hazards in Pigs and Pork (SAFEPORK), 21–24 August 2017, Foz do Iguacu, Brasil, pp. 241–244.
Bulat, A and Tzimiropoulos, G 2016. Human pose estimation via convolutional part heatmap regression. In Proceedings of the 14th European Conference on Computer Vision (ECCV), 11–14 October 2016, Amsterdam, The Netherlands, pp. 717–732.
Council of the European Union 2008. Council Directive 2008/120/EC of 18 December 2008 laying down minimum standards for the protection of pigs. Official Journal of the European Union L47, 513.
Dohoo, I, Martin, W and Stryhn, H 2012. Methods in epidemiologic research. VER Inc., Charlottetown, PEI, Canada.
European Food Safety Authority 2007. The risks associated with tail biting in pigs and possible means to reduce the need for tail docking considering the different housing and husbandry systems. The EFSA Journal 611, 113.
Fleiss, JL, Levin, B and Paik, MC 2003. The measurement of interrater agreement. In Statistical methods for rates and proportions (ed. JL Fleiss, B Levin and MC Paik), pp 598626. Wiley Interscience, Hoboken, NJ, USA.
Foddai, A, Green, LE, Mason, SA and Kaler, J 2012. Evaluating observer agreement of scoring systems for foot integrity and footrot lesions in sheep. BMC Veterinary Research 8, 65.
He, K, Zhang, X, Ren, S and Sun, J 2016. Deep residual learning for image recognition. In Proceedings of the 29th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 26 June to 1 July 2016, Las Vegas, NV, USA, pp. 770–778.
Kingma, DP and Ba, JL 2015. Adam: a method for stochastic optimization. In Proceedings of the 3rd International Conference for Learning Representations, 7–9 May 2015, San Diego, CA, USA, p. abs/1412.6980.
Kottner, J, Audigé, L, Brorson, S, Donner, A, Gajewski, BJ, Hróbjartsson, A, Roberts, C, Shoukri, M and Streiner, DL 2011. Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were proposed. Journal of Clinical Epidemiology 64, 96106.
Marzocchi, O 2014. Routine tail-docking of pigs. European Union, Brussels, Belgium.
Niedersachsen, ML 2015. Agrarminister Meyer: Ringelschwanzprämie startet mit 16,50 Euro [19.06.2015]. Retrieved on 6 February 2018 from http://www.ml.niedersachsen.de/service/pressemitteilungen/agrarminister-meyer-ringelschwanzpraemie-startet-mit-1650-euro-134624.html.
Mullan, S, Edwards, SA, Butterworth, A, Whay, HR and Main, DCJ 2011. Inter-observer reliability testing of pig welfare outcome measures proposed for inclusion within farm assurance schemes. The Veterinary Journal 190, e100e109.
Muñoz, I, Rubio-Celorio, M, Garcia-Gil, N, Guàrdia, MD and Fulladosa, E 2015. Computer image analysis as a tool for classifying marbling: a case study in dry-cured ham. Journal of Food Engineering 166, 148155.
OpenCV team 2018. Open source computer vision library 3.3. Retrieved on 6 February 2018 from https://opencv.org/
Russakovsky, O, Deng, J, Su, H, Krause, J, Satheesh, S, Ma, S, Huang, Z, Karpathy, A, Khosla, A, Bernstein, M, Berg, AC and Fei-Fei, L 2015. ImageNet large scale visual recognition challenge. International Journal of Computer Vision 115, 211252.
Simonyan, K, Vedaldi, A and Zisserman, A 2014. Deep inside convolutional networks: visualising image classification models and saliency maps. arXiv:1312.6034.
Szegedy, C, Ioffe, S, Vanhoucke, V and Alemi, AA 2017. Inception-v4, Inception-ResNet and the impact of residual connections on learning. In Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI-17), 4–9 February 2017, San Francisco, CA, USA, pp. 4278–4284.
Taylor, NR, Main, DCJ, Mendl, M and Edwards, SA 2010. Tail-biting: a new perspective. The Veterinary Journal 186, 137147.
Teixeira, DL, Harley, S, Hanlon, A, O’Connell, NE, More, SJ, Manzanilla, EG and Boyle, LA 2016. Study on the association between tail lesion score, cold carcass weight, and viscera condemnations in slaughter pigs. Frontiers in Veterinary Science 3, 24.
Vanderhasselt, RF, Sprenger, M, Duchateau, L and Tuyttens, FAM 2013. Automated assessment of footpad dermatitis in broiler chickens at the slaughter-line: evaluation and correspondence with human expert scores. Poultry Science 92, 1218.
Vieira, A, Oliveira, MD, Nunes, T and Stilwell, G 2015. Making the case for developing alternative lameness scoring systems for dairy goats. Applied Animal Behaviour Science 171, 94100.
Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

animal
  • ISSN: 1751-7311
  • EISSN: 1751-732X
  • URL: /core/journals/animal
Please enter your name
Please enter a valid email address
Who would you like to send this to? *
×

Keywords

Metrics

Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed