Side and belly lying posture detection in group pig based on binary...
1
Summary - Animal behaviour provides information about their health, welfare and environmental situations. - In different climate conditions pigs adopt different lying postures. - Visual monitoring of pig behaviours over long periods is very time consuming. - An automated machine vision technique was applied to identify side and belly lying postures of pigs. - The project took place at a commercial pig farm in Germany. - Four pens with fully slatted floor (concrete and plastic) were selected. - CCD cameras were located 2.5 meters above the ground. - Image processing algorithms in MATLAB® were developed to find each individual pig in the pens. - Support vector machine (SVM) were used for classification and detection of side and belly lying postures. Methods and Materials - Each pig was extracted from binary image and boundaries and convex hull of each animal were found. - Boundaries and convex hull value were used for training of the SVM classifier. - The trained SVM was used for detection of pigs. - Knowledge of the lying posture of each pig in the pen during lying time can be used to assess ambient conditions and improve animal welfare. - This research illustrates the application of an automated system that can detect exact lying postures of pigs. - Pigs had different postures at different thermal conditions. - Machine vision and machine learning could lead to high accuracy in pig monitoring. Conclusion Results and discussion Abozar Nasirahmadi University of Kassel Witzenhausen, 37213, Germany Email: [email protected] Phone:+495542981684 Contact - The authors wish to thank the European Pigsys project “2817ERA08D ” for the financial support. PigSys webpage: http://pigsys.eu/ Acknowledgements Side and belly lying posture detection in group pig based on binary images and SVM classifier Abozar Nasirahmadi 1* , Oliver Hensel 1 , Simone Müller 2 , Phil Kirchhofer 3 , Barbara Sturm 1,4 1- Department of Agricultural and Biosystems Engineering, University of Kassel, Witzenhausen, 37213, Germany 2- Department Animal Husbandry, Thuringian State Institute for Agriculture, Jena, 07743, Germany 3- Department of Process and Environmental Engineering, University of Applied Sciences, Konstanz, 78462, Germany 4- School of Natural and Environmental Sciences, Newcastle University, NE17RU, United Kingdom Temperature and relative humidity Ammonia Carbon dioxide Air velocity Camera Microphone Extracted RGB images Background subtraction Binary image Removing small objects Separating touching pigs Feature extraction SVM classification SVM detection SVM classifier Detection Side Belly
Transcript of Side and belly lying posture detection in group pig based on binary...
Summary- Animal behaviour provides information about their health, welfare and environmental situations.
- In different climate conditions pigs adopt different lying postures.
- Visual monitoring of pig behaviours over long periods is very time consuming.
- An automated machine vision technique was applied to identify side and belly lying postures of pigs.
- The project took place at a commercial pig farm in Germany.
- Four pens with fully slatted floor (concrete and plastic) were selected.
- CCD cameras were located 2.5 meters above the ground.
- Image processing algorithms in MATLAB® were developed to find each individual pig in the pens.
- Support vector machine (SVM) were used for classification and detection of side and belly lying postures.
Methods and Materials
- Each pig was extracted from binary image and boundaries and convex hull of each animal were found.
- Boundaries and convex hull value were used for training of the SVM classifier.
- The trained SVM was used for detection of pigs.
- Knowledge of the lying posture of each pig in the pen during lying time can be used to assess ambient conditions and improve animal welfare.
- This research illustrates the application of an automated system that can detect exact lying postures of pigs.
- Pigs had different postures at different thermal conditions.
- Machine vision and machine learning could lead to high accuracy in pig monitoring.
Conclusion
Results and discussion
Abozar NasirahmadiUniversity of Kassel Witzenhausen, 37213, GermanyEmail: [email protected] Phone:+495542981684
Contact
- The authors wish to thank the European Pigsys project “2817ERA08D ” for the financial support. PigSys webpage: http://pigsys.eu/
Acknowledgements
Side and belly lying posture detection in group pig basedon binary images and SVM classifier
Abozar Nasirahmadi1*, Oliver Hensel1, Simone Müller2, Phil Kirchhofer3, Barbara Sturm1,4
1- Department of Agricultural and Biosystems Engineering, University of Kassel, Witzenhausen, 37213, Germany
2- Department Animal Husbandry, Thuringian State Institute for Agriculture, Jena, 07743, Germany
3- Department of Process and Environmental Engineering, University of Applied Sciences, Konstanz, 78462, Germany
4- School of Natural and Environmental Sciences, Newcastle University, NE17RU, United Kingdom
Temperature and relative humidity Ammonia Carbon dioxide
Air velocity CameraMicrophone
Extracted RGB images Background subtraction Binary image Removing small objects
Separating touching pigsFeature extraction SVM classificationSVM detection
SVM classifier Detection
Side
Belly
