Convolutional neural network (CNN)-based computer vision systems have been increasingly applied in animal farming to improve animal management, but current knowledge, practices, limitations, and solutions of the applications remain to be expanded and explored. The objective of this study is to systematically review applications of CNN-based computer vision systems on animal farming in terms of the five deep learning computer vision tasks: image classification, object detection, semantic/instance segmentation, pose estimation, and tracking. Cattle, sheep/goats, pigs, and poultry were the major farm animal species of concern. In this research, preparations for system development, including camera settings, inclusion of variations for data recordings, choices of graphics processing units, image preprocessing, and data labeling were summarized. CNN architectures were reviewed based on the computer vision tasks in animal farming. Strategies of algorithm development included distribution of development data, data augmentation, hyperparameter tuning, and selection of evaluation metrics. Judgment of model performance and performance based on architectures were discussed. Besides practices in optimizing CNN-based computer vision systems, system applications were also organized based on year, country, animal species, and purposes. Finally, recommendations on future research were provided to develop and improve CNN-based computer vision systems for improved welfare, environment, engineering, genetics, and management of farm animals.

No abstract

Since its recognition in the early 1980s, enteric septicemia of catfish (ESC), caused by the gram-negative enteric Edwardsiella ictaluri, is one of the most significant pathogens affecting catfish aquaculture. In efforts to improve disease management in U.S. catfish aquaculture, a live-attenuated ESC vaccine and method of in-pond oral vaccine delivery was developed. This work evaluated the efficacy of the oral delivery platform using a live-attenuated E. ictaluri vaccine under simulated commercial conditions. Channel catfish fingerlings were orally vaccinated approximately 40-50 days poststocking by mixing the attenuated vaccine with feed using a patent-pending mechanized delivery system. Across three discrete years under varying evaluation methods, vaccination resulted in significant improvements in survival, feed conversion ratio, feed fed, and total yield. An economic analysis of experimental data suggests a significant positive net economic benefit from vaccination.After accounting for a projected cost of vaccination, the realized improvements in survival and yield resulted in an estimated benefit exceeding $3,500/ha ($1,400/acre) on fingerling operations. These results suggest vaccinating channel catfish fingerlings against ESC is an effective strategy to mitigate ESC-related losses and can significantly improve production efficiency and profitability on catfish operations.

No abstract

HighlightsUsing an elevated platform with a manure catcher reduced litter moisture content and ammonia concentration.A robotic vehicle encouraged bird movement and use of the elevated platform.Using the elevated platform and robotic vehicle jointly improved broiler paw quality and plumage cleanliness.Abstract. Elevated platform (EP) and robotic vehicle (RV) are two emerging systems aiming to improve environment enrichment and bird activity in broiler housing systems. However, the impacts of these systems on broiler production, welfare, and housing environment have not been evaluated. The objective of this study was to evaluate the effects of using EP and RV individually or jointly on broiler feed conversion ratio (FCR), litter moisture content (LMC), ammonia (NH3) concentration, gait score (GS), paw quality (PQ), plumage cleanliness (PC), and bird activity index (AI). Broilers were reared for eight-week production cycles. Four experimental rooms (54 birds room-1) were randomly assigned four treatments: EP only, RV only, EP and RV (EP+RV), and neither EP nor RV (Ctrl). Broiler GS, PQ, and PC were assessed following welfare protocols. Bird AI was determined through image processing. The experiment was repeated three times. The results showed that overall FCR values were 1.806 for EP, 1.804 for RV, 1.797 for EP+RV, and 1.791 for Ctrl. The normalized LMC values were 23.2% ±4.1% for EP, 32.8% ±4.1% for RV, 23.4% ±2.5% for EP+RV, and 35.7% ±7.0% for Ctrl over the eight-week production cycles. NH3 concentrations were 40% lower in the rooms with EP than in the rooms without EP at the end of the production cycle. Broilers had better PQ in the rooms with EP than in the rooms without EP. Broiler PC seemed better in the rooms with RV compared to those without RV. Operation of RV increased bird AI; however, the benefits in activity encouragement diminished as the broilers grew. The number of broilers on the EP was higher in the EP+RV rooms than in the EP rooms. It is concluded that using EP and RV together may improve broiler welfare and activity without compromising their production performance. Keywords: Behavior, Broiler, Elevated platform, Robotic vehicle, Welfare.