Source: UNIV OF WISCONSIN submitted to NRP
DEVELOPMENT OF AN AUTOMATED COMPUTER VISION SYSTEM TO MONITOR BEHAVIOR OF PRE-WEANED DAIRY CALVES
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
HATCH
Reporting Frequency
Annual
Accession No.
1014469
Grant No.
(N/A)
Cumulative Award Amt.
(N/A)
Proposal No.
(N/A)
Multistate No.
(N/A)
Project Start Date
Oct 26, 2017
Project End Date
Sep 30, 2021
Grant Year
(N/A)
Program Code
[(N/A)]- (N/A)
Recipient Organization
UNIV OF WISCONSIN
21 N PARK ST STE 6401
MADISON,WI 53715-1218
Performing Department
Animal Sciences
Non Technical Summary
Calfhood diseases can adversely affect productive performance during early stages of growth as well as overall lifetime performance of dairy cows. Such negative impact of calf diseases on productive performance results in considerable economic losses in the dairy industry. Calves dramatically alter their behavior when they are sick, and even before getting sick, so that behavior pattern could be used as an earlier indicator of health or management problems. However, in large dairy operations, the daily monitoring of calf behavior becomes laborious, and the large number of animals becomes a limiting factor for such evaluation. In this context, an efficient use of calf behavior for early detection of healthy issues in commercial farms is unfeasible. The overall objective of this project is to develop an automated computer vision system to monitor behavior in pre-weaned calves through image analysis. The computer vision system will comprise an integrated computational process with image acquisition using cameras in the field, upload of information to a cloud computing service, and real-time data analytics. Results from this project will provide an important computational system to predict dairy calf behavior through automated image analyses, which can be then applied in commercial herds to anticipate potential health problems and guide preventive management practices.
Animal Health Component
30%
Research Effort Categories
Basic
50%
Applied
30%
Developmental
20%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
3073410208050%
3153410209050%
Knowledge Area
307 - Animal Management Systems; 315 - Animal Welfare/Well-Being and Protection;

Subject Of Investigation
3410 - Dairy cattle, live animal;

Field Of Science
2090 - Statistics, econometrics, and biometrics; 2080 - Mathematics and computer sciences;
Goals / Objectives
The overall objective of this project is to develop an automated computer vision system to monitor behavior in pre-weaned calves through image analysis. The computer vision system will comprise an integrated computational process with image acquisition in the field, upload of information to a cloud computing service, and real-time data analytics. Three specific aims will be pursued: a) Development and improvement of the logistics pertaining to hardware network and software tools to implement the computer vision system and all its related components: regular and infrared cameras, Wi-Fi connection, cloud computing, data storage and security, b) Development of data analysis tools for automated calf identification based on dorsal image information, and c) Use of deep learning algorithms to predict behavior of individual and group animals housed in hutches or group pen.
Project Methods
Two different facilities will be used to test the computer vision system capability: The Dairy Cattle Research Center (DCRC) and the Emmons Blaine Dairy Cattle Research Center (Arlington-WI), of the University of Wisconsin-Madison. Calves with be housed either as a group pen, or in individual hutches. Regular Wi-Fi cameras (Amcrest Outdoor Wi-Fi Camera, model: IP3M-956EW) will be installed 5 meters above the hutches, being one camera per hutch. In addition, Kinect cameras (Microsoft Xbox one Kinect sensor, model: 2.0) will be installed above the hutches in a mobile support, which will allow us to move the equipment among hutches. The calves will be monitored by the Kinect cameras every 2 days during 24 hours, thus the 10 cameras will switch over between the 2 groups of 10 animals (hutches) every 2 days. The video stream data will be transferred from the camera to a computer server using Wi-Fi connection, and later uploaded to the cloud for computations.The animals housed as group pen will be classified as one of the following behavior parameters: lying down, eating, drinking, walking/standing. The behavior parameters for calves housed in individual hutches will be: time spent inside of the hutch, time spent outside of the hutch, time spent drinking milk, and time spent outside of the hutch and not drinking milk when milk is fed. Those parameters will be defined through visual observations by using the videos recorded, and those will be used as the observed data (standard) for supervised computational learning as described below.The images stored in the blob storages will be generated at 1 fps video slice rate. As way to define the optimum number of frames to be exported, we will train the computer algorithm using 5 different rates: 1 frame per second (standard), 1 frame every 5 seconds, 1 frame every 10 seconds, 1 frame per every 20 seconds, and 1 frame every 30 seconds. This strategy may reduce the number of frames to be sent to the cloud from 86,400 (1 fps) to 2,880 (1 frame every 30 seconds) per hutch or pen, which can dramatically reduce the computational burden.We will use Convolutional Neural Network (CNN) as a machine learning classifier (Abadi et al., 2015). A CNN is a deep network topology that can be formed as a computational graph, and Google's open-source computational platform, TensorFlow, is directly suited for this application (Abadi et al., 2015). Keras is a Phyton library containing numerous implementations of commonly used neural network building blocks such as layers,objectives,activation functions,optimizers, and a host of tools to make working with image and text data easier. Keras is a high level API built on TensorFlow and it is a friendly and easy to use interface to work with the CNN architectures. In both experiments we will use Keras and TensorFlow for the CNN definition, training and classification. This platform has the advantage of using Graphical Processing Unit (GPU) to speed data processing, which it will be used in our study.The hyperparemeters such as number of layers, activation function, optimizers, learning rate, and others will be defined during the model training phase, since our main goal is to maximize the accuracy to classify a specific activity. In addition, most of those parameters are very data-type dependent, which means that specific hyperparmeters can be adequate for a set of images/data but not for others.Images from the first week of both trials will be manually separated in folders related to its respective behavior. For example, from the total of 86,400 daily pictures of 1 calf/hutch, the pictures will be split into 4 categories (folders): 1) inside of the hutch; 2) outside of the hutch; 3) drinking milk when milk is fed; and 4) not drinking milk when milk is fed. Those pictures will be used to train and validate the CNN model. The images of the following 3 weeks will be used as a test set, and the models previously developed will be evaluated. The same approach will be used in the experiment with group pen calves, for which 4 categories (folders) will be created to classify images from the first week of trial: 1) eating; 2) drinking; 3) lying down, and 4) walking/standing. However, an additional model to predict the animal identification (ID) will be developed for the pen group trial. As animals will be housed in group in this case, we will first develop a model to predict the animal ID based on the dorsal images, considering that each Holstein calf will present a unique colt color/pattern. Thus, the model will identify the calf, and then it will classify each individual calf's behavior.To train the CNN model, a computer coupled with a GPU processor (EVGA GeForce GTX 1080) will be used. After trained and validated, the models will be uploaded to the cloud service to automatically process the images that will constantly arrive in the blob storage. The cloud service will be set to generate behavior predictions in 3 different timelines: 1) Three times a day: 6 a.m., 12 p.m., and 6 p.m.; 2) Twice a day: 6 a.m and 6 p.m; and 3) Once a day: 6 p.m. The main goal of these three different timelines is to evaluate the frequency needed to predict behavior in order to maximize the decision-making at the farm level with the lowest computational cost/burden.

Progress 10/26/17 to 09/30/21

Outputs
Target Audience:The proposed computer vision system to monitor cattle behavior and development will be of interest to researchers working on nutrition and production, as it might be used to assist on data collection and cattle management in research experiments. More importantly, the system will be extremely useful for extensionists and cattle producers as a toll to help making optimal management decisions. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project supported a graduate student (Mr. Arthur Fernandes). In addition, during the reported period, the project has attracted an international visiting scholar and a visiting scientist to the University of Wisconsin-Madison: Alexandre Cominotte, Visiting Scholar from the Sao Paulo State University, Brazil; Talita Santana from the Federal University of Vicosa, Brazil; and Dr. Thais Basso, Visiting Scientist from Embrapa - Brazil. These researchers were funded by their own institutions. How have the results been disseminated to communities of interest?Results from this project have been disseminated to communities of interest via peer-reviewed publications, scientific meeting presentations/abstracts, invited talks and seminars. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Image analysis and computer vision can be used as a powerful tool to monitor animal behavior and animal growth and development. The overall objective of this project was todevelop an automated computer vision system (CVS) to monitor behavior in dairy cattle through image analysis.The developed CVS comprises an integrated computational process with image acquisition in the field, upload of information to a cloud computing service, and real-time data analytics. Images are automatically captured from unrestrained animals and transferred to a central computer and to the cloud. Images are then processed and subjected to object detection, i.e. identification of animals on each frame, and classification of their behavior, e.g. standing, lying, eating, and drinking. A body weight prediction model was also developed, which includes an additional step of image segmentation to select pixels that pertain to animal body as opposed to background, from which biological features of interested are obtained, e.g. top view area of the body, length, width, perimeter, etc. Such features are then used on a predictive model to estimate body weight of each animal.

Publications

  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Rosa, G. J. M. Grand challenge in precision livestock farming. Frontiers in Animal Science 2: 650324, 2021.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Silva, F. F., Morota, G. and Rosa, G. J. M. High-throughput phenotyping in the genomic improvement of livestock. Frontiers in Genetics 12: 707343, 2021.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Ribeiro, L. A. C., Bresolin, T., Rosa, G. J. M., Casagrande, D. M., Danes, M. A. C. and D�rea, J. R. R. Disentangling data dependency using cross-validation strategies to evaluate prediction quality of cattle grazing activities using machine learning algorithms and wearable sensor data. Journal of Animal Science 99(9): 1-8, 2021.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Rosa, G. J. M. Combining Big Data Analytics and Omics Techniques to Improve Beef Cattle Selection and Production. II International Livestock Studies Congress. Antalya, Turkey. Oct 29-30, 2021.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Rosa, G. J. M. Data Mining in Livestock Production: Artificial and Natural Intelligence Working Together. XIII Brazilian Congress on Agroinformatics. Nov 10-12, 2021. Virtual Meeting.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Rosa, G. J. M. Combining Big Data Analytics and Omics Techniques to Improve Beef Cattle Selection and Production. 5th International Meeting on Plant Breeding. Piracicaba, Brazil. Oct 5-6, 2021.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Rosa, G. J. M. Computer Vision in Beef Cattle Production. Research and Technological Innovations for a Sustainable Cattle Farming, Virtual International Symposium, National Institute of Agricultural Research (INIA), Peru. June 30-July 2, 2021.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Souza, F. M., Lopes, F. B., Rosa, G. J. M. and Magnabosco, C. U. Economic values of reproductive, growth, feed efficiency and carcass traits in Nellore cattle. Journal of Animal Breeding and Genetics 139: 170-180, 2021.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Moreira, L. C., Rosa, G. J. M. and Schaefer, D. M. Board Invited Review: Beef production from cull dairy cows: a review from culling to consumption. Journal of Animal Science 99(7): 1-18, 2021.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Moreira, L. C., Passafaro, T. L., Schaefer, D. M. and Rosa, G. J. M. The effect of life history events on carcass merit and price of cull dairy cows. Journal of Animal Science 99(1): skaa401, 2021.


Progress 10/01/19 to 09/30/20

Outputs
Target Audience:The proposed computer vision system to monitor behavior of pre-weaned dairy calves will be of interest to researchers working on dairy calf growth, development and behavior, as it might be used to assist on data collection and calf management in research experiments. More importantly, the system will be extremely useful for extensionists and dairy producers as a toll to help making management and culling decisions during dairy calves' growth. Changes/Problems:COVID-19 certainly had an effect on our project, limiting us to access animals at the UW farms, and also limiting the interaction among project participants. Nonetheless, we were still able to satisfactorily progress with our project goals and everything is on target to finish on time. What opportunities for training and professional development has the project provided?The project funded a graduate student (Mr. Arthur Fernandes). In addition, during the reported period, the project has attracted an international visiting scholar and a visiting scientist to the University of Wisconsin-Madison: Alexandre Cominotte, Visiting Scholar from the Sao Paulo State University, Brazil; and Dr. Thais Basso, Visiting Scientist from Embrapa - Brazil. These researchers were funded by their own institutions. How have the results been disseminated to communities of interest?Preliminary results from this project have already been disseminated to communities of interest via peer-reviewed publications, scientific meeting presentations/abstracts, invited talks and seminars. What do you plan to do during the next reporting period to accomplish the goals?The next period of the project will be devoted to final data analysis steps for improvement of algorithms. Results from this project will provide an important computational system to predict dairy calf behavior through automated image analyses, which can be then applied in commercial herds to anticipate potential health problems, and guide preventive management practices of dairy calves.

Impacts
What was accomplished under these goals? Image analysis and computer vision can be used as a powerful tool to monitor animal behavior and animal growth and development. During the third year of this project, we developed and compared different computer vision models and algorithms to detect animals on images and predict their body weight and body composition. Two strategies were considered: 1) a two-step approach with image preprocessing followed by prediction model, and 2) prediction using raw images as input. The image preprocessing step included image segmentation and extraction of image features, such as area, length, width, and volume. Prediction models included traditional multiple linear regression and partial least squares regression, as well as ridge regression and lasso. For the single step approach, a deep learning (DL) image encoders approach was employed. The techniques were applied not only to cattle, but also to pigs and fish. The DL method achieved the best overall performance, with the lowest mean absolute scaled error (MASE) and root mean square error, and the highest predictive squared correlation (R2). For example, with a study using 12,000 images from 557 finishing pigs, with body weight and body composition traits, a cross-validation assessment of model performance indicated MASE of 2.69, 5.02, and 13.56, and R2 of 0.86, 0.50, and 0.45, for body weight, muscle depth, and back fat, respectively. In summary, results show that image analysis provide a high accuracy prediction of body weight and composition, especially algorithms based on deep neural networks.

Publications

  • Type: Journal Articles Status: Accepted Year Published: 2021 Citation: Rosa, G. J. M. Grand challenge in precision livestock farming. Frontiers in Animal Science (accepted)
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Fernandes, A. F. A., Dorea, J. R. R., Valente, B. D., Fitzgerald, R., Herring, W. and Rosa, G. J. M. Comparison of data analytics strategies in computer vision systems to predict pig body composition traits from 3D images. Journal of Animal Science 98(8): skaa250, 2020.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Fernandes, A. F. A., D�rea. J. R. R. and Rosa, G. J. M. Image analysis and computer vision applications in animal sciences: an overview. Front. Vet. Sci. 7: 551269, 2020.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Fernandes, A. F. A., Turra, E. M., Alvarenga, E. R., Passafaro, T. L., Lopes, F. B., Alves, G. F. O., Singh, V. and Rosa, G. J. M. Deep Learning image segmentation for extraction of fish body measurements and prediction of body weight and carcass traits in Nile tilapia. Computers and Electronics in Agriculture 170: 105274, 2020.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Cominotte, A., Fernandes, A. F. A., Dorea, J. R. R., Rosa, G. J. M., Ladeira, M. M., van Cleeff, E. H. C. B., Pereira, G. L., Baldassinic, W. A. and Machado Neto, O. Automated computer vision system to predict body weight and average daily gain in beef cattle during growing and finishing phases. Livestock Science 232: 103904, 2020.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2020 Citation: Rosa, G. J. M. RFID-Based Animal Identification and Traceability for Cull Dairy Cows. 2020 Cattle Industry Convention and NCBA Trade Show, San Antonio, TX. February 5-7, 2020.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2020 Citation: Rosa, G. J. M. Advancing Livestock Genetic Improvement with High-Throughput Phenotyping Approaches. XXVIII Plant & Animal Genomes (PAG) Conference, San Diego, CA. January 11-15, 2020.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2020 Citation: Ferreira, R. E. P., Pereira, L. G. R., Bresolin, T., Rosa, G. J. M. and Dorea, J. R. R. Development of an identification system to recognize individual animals based on biometric facial features. Journal of Dairy Science 103(Supplement 1): 240, 2020.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2020 Citation: Rosa, G. J. M., Aiken, V. C., Fernandes, A. and Dorea, J. R. Enhancing Beef Production and Quality Using Big Data Analytics and Computer Vision Techniques. ASAS Midwest Section / ADSA Midwest Branch 2020 Joint Meeting, Omaha, NE. March 2-4, 2020.


Progress 10/01/18 to 09/30/19

Outputs
Target Audience:The proposed computer vision system to monitor behavior of pre-weaned dairy calves will be of interest to researchers working on dairy calf growth, development and behavior, as it might be used to assist on data collection and calf management in research experiments. More importantly, the system will be extremely useful for extensionists and dairy producers as a toll to help making management and culling decisions during dairy calves' growth. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project supported a graduate student (Mr. Alexandre Cominotte). In addition, during the reported period, the project has attracted an international visiting scholar and a visiting scientist to the University of Wisconsin-Madison: Tiago Bresolin, Visiting Scholar from the Sao Paulo State University, Brazil; and Dr. Yangfan Wang, Visiting Scientist from the Ocean University of China - China. These researchers were funded by their own institutions. How have the results been disseminated to communities of interest?Preliminary results from this project have already been disseminated to communities of interest via scientific meeting presentations/abstracts, invited talks and seminars. What do you plan to do during the next reporting period to accomplish the goals?The next period of the project will be devoted to further development of the logistics pertaining to hardware network and software tools, additional data analysis, as well as additional trials to increase sample size and increase prediction accuracy of the system. Results from this project will provide an important computational system to predict dairy calf behavior through automated image analyses, which can be then applied in commercial herds to anticipate potential health problems, and guide preventive management practices of dairy calves.

Impacts
What was accomplished under these goals? During the second year of the project, we extend our methods for tracking a large number of animals, by individually identifying each animal in real time as well as monitoring their behavior. The calves' behavior was classified into four categories: lying, standing, eating, and drinking. Preliminary results have been presented at a scientific meeting. Another trial was carried out at the Dairy Cattle Research Center at the University of Wisconsin-Madison to test the improved system. In this study, five animals were housed as a group, and a regular Wi-Fi camera was installed at the top of the pen. The pipeline involving the data transfer, storage and security was successfully evaluated. Our system tracks calves based on their coat color patterns, and simultaneously assess their behavior. The computer vision system implemented in this study showed high accuracy values to identify each calf (between 70 and 92%), as well as classify their behavior (between 85 and 100%).

Publications

  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Fernandes, A. F. A., Dorea, J. R. R., Fitzgerald, R., Herring, W. and Rosa, G. J. M. A novel automated system to acquire biometric and morphological measurements, and predict body weight of pigs via 3D computer vision. Journal of Animal Science 97: 496-508, 2019.
  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Koltes, J. E., Cole, J. B., Clemmens, R., Dilger, R. N., Kramer, L. M., Lunney, J. K., McCue, M. E., McKay, S. D., Mateescu, R. G., Murdoch, B. M., Reuter, R., Rexroad, C.E., Rosa, G. J. M., Ser�o, N. V. L., White, S. N., Woodward-Greene, M. J., Worku, M., Zhang, H. and Reecy, J. M. A Vision for development and utilization of high-throughput phenotyping and big data analytics in livestock. Frontiers in Genetics 10: 1197, 2019.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2019 Citation: Rosa, G. J. M. Opportunities and Challenges in Statistical Analysis of High-Throughput Phenotyping Data in Livestock. In: 64th Annual Meeting of the Brazilian Region (RBRAS) of the International Biometric Society, Cuiab�, MT - Brazil. July 29 - Aug 02, 2019.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2019 Citation: Rosa, G. J. M., Dorea, J. R. R., Fernandes, A. F. A., and Passafaro, T. Leveraging on High-Throughput Phenotyping Technologies to Optimize Livestock Genetic Improvement and Husbandry. In: 2019 ASAS-CSAS Annual Meeting and Trade Show. Austin, TX. July 8-11, 2019.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2019 Citation: Rosa, G. J. M., Dorea, J. R. R., Fernandes, A. F. A., and Passafaro, T. Leveraging on High-Throughput Phenotyping Technologies to Optimize Livestock Genetic Improvement and Husbandry. In: 23rd Congress of the Animal Science and Production Association (ASPA). Sorrento - Italy, June 11-14, 2019.


Progress 10/26/17 to 09/30/18

Outputs
Target Audience:The proposed computer vision system to monitor behavior of pre-weaned dairy calves will be of interest to researchers working on dairy calf growth, development and behavior, as it might be used to assist on data collection and calf management in research experiments. More importantly, the system will be extremely useful for extensionists and dairy producers as a toll to help making management and culling decisions during dairy calves' growth. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project supports a graduate student (Mr. Alexandre Cominotte). In addition, the project has attracted international visiting scholars to the University of Wisconsin-Madison: Dr. Tatsuhiko Goto, from the Obihiro University of Agriculture and Veterinary Medicine in Japan; and Dr. Otavio Machado Neto, from the Sao Paulo State University, Brazil. How have the results been disseminated to communities of interest?Preliminary results from this project have already been disseminated to communities of interest via scientific meeting presentations/abstracts, invited talks and seminars. The presentation was at the 69th Annual Meeting of the European Federation of Animal Science (EAAP). Dubrovnik - Croacia, August 27-31, 2018. What do you plan to do during the next reporting period to accomplish the goals?The next period of the project will be devoted to further development of the logistics pertaining to hardware network and software tools, additional data analysis, as well as additional trials to increase sample size and increase prediction accuracy of the system.

Impacts
What was accomplished under these goals? During this first year of the project, we performed extensive literature review, and tested different equipment such as cameras, Wi-Fi connection and cloud computing alternatives. A pilot study has been already carried out at the Dairy Cattle Research Center at the University of Wisconsin-Madison to test the system. In this pilot study, two animals were housed in individual pens, and one regular Wi-Fi camera was installed at the top of the pen, and during the 12 days of trial a video stream was recorded. The pipeline involving the data transfer, storage and security was successfully evaluated. Our system tracks the objects (calves) across frames (which means across time), and then predict the behavior parameters, such as eating, drinking, walking/standing, and lying down. Briefly, the video stream was sliced in 1 frame per second (fps), and the image background was discounted of all images, which resulted only in the objects to be tracked. Thus, we tracked how the calves moved into the frames across time, and from this tracking we calculated the total distanced traveled, and time spent walking/standing, lying, eating and drinking. The image analysis implemented in this pilot study showed high accuracy values (66 to 98%) to predict behavior depending on the type of behavior. Results from this project will provide an important computational system to predict dairy calf behavior through automated image analyses, which can be then applied in commercial herds to anticipate potential health problems, and guide preventive management practices of dairy calves.

Publications

  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Rosa, G. J. M., Dorea, J. R. R., Fernandes, A. F. A., Ferreira, V. C. and Passafaro, T. L. Leveraging on high-throughput phenotyping technologies to optimize livestock genetic improvement and husbandry. In: 69th Annual Meeting of the European Federation of Animal Science, Abstract 349, Dubrovnik, Croatia, August 27-31, 2018.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Dorea, J. R. R., A. F. A. Fernandes, V. C. Ferreira, D. K. Combs, and G. J. M. Rosa. Use of 3-dimensional camera to predict body weight in pre-weaned dairy calves. In: 2018 Annual Meeting of American Society of Dairy Science, 2018, Knoxville-TN.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Cominotte, A., A. F. A Fernandes, M. M. Ladeira, M. L. Chizzotti, J. R. R. Dorea, O. R. Machado Neto, and G. J. M. Rosa. Prediction of average daily gain in beef cattle utilizing 3D images. In: 2018 Annual Meeting of Brazilian American Society of Animal Science, 2018, Goiania-GO, Brazil. 2018.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Cominotte, A., A. F. A Fernandes, M. M. Ladeira, M. L. Chizzotti, J. R. R. Dorea, O. R. Machado Neto, and G. J. M. Rosa. Prediction of body weight and animal growth in beef cattle utilizing 3D images. In: 2018 Annual Meeting of Brazilian American Society of Animal Science, 2018, Goiania-GO, Brazil. 2018.