About

Angela Green-Miller is associated with the Center for Digital Agriculture at the University of Illinois. The center focuses on research and development in digital and precision agriculture, including AI-driven tools, data collection, and analysis to optimize agricultural practices. The center offers interdisciplinary programs such as a Master’s Degree in Engineering with a concentration in Digital Agriculture, and collaborates internationally through initiatives like the Global Perspectives on Digital/Smart Agriculture Seminar Series. The Center for Digital Agriculture is engaged in projects like AI AgriBench, CropWizard, and the PigLife Dataset, which aim to build trust in AI applications in agronomy, provide decision-support services for agricultural professionals, and develop benchmarks for computer vision systems in livestock industries. The center also hosts events such as conferences, hackathons, and seminars to promote innovation and knowledge exchange in digital agriculture. Its efforts include supporting research, education, and industry partnerships to advance the adoption of AI and digital technologies in agriculture.

Research topics

• Computer Science
• Biology
• Animal science
• Ecology
• Mathematics
• Business
• Data science
• Engineering
• Psychology
• Systems engineering
• Risk analysis (engineering)
• Statistics
• Physics

Selected publications

• PSIII-19 Identification of important behaviors and postures to indicate respiratory distress in pigs.

  Journal of Animal Science · 2025-10-01 · 1 citations

  articleOpen accessSenior author

  Abstract Respiratory illness is a prevalent issue in pork production systems, resulting in economic losses each year. Behavioral analysis may detect early onset of respiratory illness in breeding herds, increasing the overall health and welfare of the animals. The objective of this study was to determine if respiratory distress could be characterized by behavior and posture differences for use as a tool to identify emerging subclinical respiratory illness in late-gestation pigs. In this study, a cohort of 10 gestational sows were housed in individual stalls, fed a diet of corn-soybean meal, and provided ad libitum access to water. Animals were randomly split between two main treatment groups: a control group (10 μg phosphate buffered saline/kg BW) and immune challenge group (10 μg lipopolysaccharide (LPS)/kg BW). Treatment was administered intravenously (jugular venipuncture) on gestational day 70. The pigs were recorded continuously using video monitoring. Three unexpected mortalities were observed. Necropsy performed by veterinary staff determined the mortalities resulted from an undetected subclinical respiratory illness. An ethogram containing 27 behavior and posture labels was created to study the activity of the sows. Behavioral Observation Research Interactive Software, or BORIS, was used to analyze the time frames 24-hour hours pre-LPS injection; and 0.5-, 1.5-, and 2-hours post-LPS injection. The analysis was split into 3 treatment groups: control, LPS+Recover, and LPS+Death. A statistical analysis was performed to compare the behavior data between each treatment group. Normal and transformed data was analyzed using an ANOVA comparison, and non-normal data was analyzed using a Kruskal-Wallis rank sum test. Differences in feeding (P=0.0131), drinking (P=0.0218), sham chewing (P=0.0131), and head shaking (P=0.011) behaviors were found between the control group and both LPS treatment groups for both average bout length and total number of occurrences. Each of these behaviors occurred less within the LPS+Recover and LPS+Death groups with all four behaviors occurring less for LPS groups. These results may indicate that these behaviors may be useful to monitor as sentinel behaviors for health status changes and early detection of infections and/or disease. In the video, some sows demonstrated convulsive-like behaviors, which have not been studied extensively before in previous literature. Further analysis is needed to determine its impact on sickness behavior, specifically respiratory distress. Results from this study may be useful in behavior classifications of gestational sows responding to LPS treatments and for animal health monitoring.

  PublisherOA PDFDOI

• 298 Impact of ruminal acidosis on cattle energy metabolism

  Journal of Animal Science · 2024-05-01 · 1 citations

  articleOpen access

  Abstract The objective of this study was to determine the impact of a bout of ruminal acidosis on energy metabolism of cattle unadapted to a high concentrate diet. Eleven ruminally cannulated steers [body weight (BW) = 352 kg ± 27] were blocked into 3 groups based on initial BW. Before the start of the experiment, animals were consuming a forage-based diet as well as adapted to the headbox style respiration chambers. Additionally, before the experiment, gas emission data were collected over a 24-h period when cattle received an ad libitum forage-based diet for use as a covariate in the statistical analysis. For the experiment, steers were moved into headboxes at the conclusion of a 24-h fast and subsequently received 1 of 2 treatment diets: control (CON), forage-based diet or acidosis (AC), concentrate-based diet. Steers remained in the headboxes for 48 h. Gas concentrations from each headbox were collected hourly and analyzed with an infrared photoacoustic gas analyzer. Ruminal pH, fecal pH, volatile fatty acids, and lactate were also measured during the acidosis challenge. Data were analyzed with the MIXED procedure of SAS 9.4. A treatment × day effect (P = 0.03) was observed for dry matter intake with intake being similar for both CON and AC steers on d 1 but AC steers consuming 2.02 kg less on d 2. Treatment affected ruminal pH (P < 0.01) as CON steers had a greater ruminal pH than AC steers. Greater total volatile fatty acids (P < 0.01) were observed for steers on AC treatment compared with CON. A treatment × time interaction (P < 0.01) was observed for ruminal lactate concentration with AC steers having greater concentrations from h 16 to 36. Fecal pH was affected by a treatment × time interaction (P = 0.05) as AC steers had lower fecal pH from h 24 to h 32. Over the 48-h observation period, there was a treatment × time interaction (P = 0.04) for carbon dioxide production as AC steers produced less CO2 after h 24. No treatment × time interaction was detected for oxygen consumption. However, there was a time effect (P < 0.01). Respiratory quotient was not affected by treatment, day, or their interaction (P ≥ 0.19). Heat production tended (P = 0.06) to be different between treatments, with AC having less heat production than CON. Thus, the effects of ruminal acidosis on gas emissions and energetics may be more transient during the time course of an acidotic bout or have a longer duration even after ruminal pH has recovered.

  PublisherOA PDFDOI

• AIFARMS: Artificial intelligence for future agricultural resilience, management, and sustainability

  AI Magazine · 2024-02-22 · 12 citations

  articleOpen access

  Abstract The AIFARMS Artificial Intelligence for Future Agricultural Resilience, Management, and Sustainability national AI institute brings together over 40 world‐class AI and agriculture researchers, with the common mission to develop foundational advances in AI and use them to ensure that future agriculture is environmentally friendly, sustainable, affordable, and accessible to diverse farming communities. Since its establishment in 2020, AIFARMS has advanced the state of the art in autonomous farming, cover crop planting, machine learning for improved outcomes from remote sensing, dynamic estimation of yield loss from weeds, and livestock management. The institute has prioritized the creation and utilization of high‐quality, openly available data sets for advancing foundational AI and tackling agricultural challenges. AIFARMS leverages a close partnership between UIUC and Tuskegee University to build programming for a skilled and diverse next‐generation workforce in digital agriculture. We are expanding the reach of AIFARMS outside of the current partners to collaborate with national AI institutions and international partners.

  PublisherOA PDFDOI

• 210 Sleeping beauties: A daily time budget for individually housed research pigs at 6-d of age

  Journal of Animal Science · 2024-05-01

  articleOpen accessSenior author

  Abstract The pre-wean period of development is difficult to manage and can impact the success of subsequent production stages. Behavior observations can provide insight into the health and welfare status of animals and allow for efficient and timely management interventions. A validated time budget can be utilized as a baseline for further comparisons and the detection of changes in behavior over time. In this project, 24-h of video data for 21 individually housed research pigs at 6 d of age was evaluated using a validated behavior ethogram. The behavior ethogram contained 5 main categories (inactive, consumption, exploratory, social, and other) and 23 behavior labels. The ethogram was applied using a validated sampling strategy, which included 5-min continuous sampling intervals at the start of each hour for active behaviors and 5-min instantaneous sampling intervals for inactive behaviors. The data from each behavior observation was quantified into durations and presented as percentage of observations spent performing a specific behavior label within the observation period. The daily time budget revealed the following time occupation across the 5 main behavior categories: inactive, 98.60% (lying, 85.43%; not lying, 13.14%); consumption, 0.30%; exploratory, 0.97%; social, 0.01%, and other, 0.12%. As expected for very young pigs, most of the day was inactive, with almost 21 h lying. The most performed behavior was “lying socially”, and this behavior represented 38.48% of the overall daily time budget. The remaining 46.95% of time spent lying was spent “lying sternal”, “lying lateral”, or “lying on an object” in the pen. The pigs are housed individually, but they are able to see, hear, and smell a neighboring pig through a plexiglass barrier. The behavior observation “lying socially” is defined as follows: “The piglet is lying (sternal or lateral) touching the plexiglass barrier. The piglet in the adjacent pen is lying directly on the other side of the barrier, and the piglets would be touching each other if the barrier was not in place.” This observation provides insight into the importance of social resting for young pigs.

  PublisherOA PDFDOI

• Promote computer vision applications in pig farming scenarios: High-quality dataset, fundamental models, and comparable performance

  Journal of Integrative Agriculture · 2024-08-23 · 12 citations

  articleOpen accessSenior author

  Computer vision is widely recognized as an influential technology in the field of precision management of animals. Emerging studies have demonstrated the potential to improve pig health and welfare through animal surveillance systems and computer vision (CV) algorithms. However, the lack of benchmark datasets and robust fundamental algorithms restrict CV applications for the commercial use. This study aims to bridge the gap between technology development and commercial applications in pig farming scenarios by introducing a general-purpose dataset ( PigLife ), comparing benchmark performances of foundational CV algorithms and model development workflows. The PigLife dataset contains video clips and images (38 short video clips, 2K image frames, 22K pig instances) across most pig production phases in a typical commercial pig farm: Breeding and Gestation, Farrow to Wean, Weaning & Nursery, and Growth to Finish. Three detection algorithms ( Faster R-CNN , RetinaNet , TridentNet ) and three segmentation algorithms ( Mask R-CNN , MViTv2 , Point-Rend ) were trained on the PigLife dataset from scratch. Fine-tuning of pre-trained models ( YOLO8-m , Faster-RCNN-r50 ) and no-training from zero-shot models ( CLIP-SAM , Grouddino-HQSAM ) were also evaluated to suggest faster CV development workflows for commercial applications in pig farming. This study emphasizes the necessity of a benchmark dataset for evaluating the robustness of algorithms and identifying the remaining difficulties and challenges across various algorithms. Furthermore, developing CV models from pre-trained algorithms or zero-shot models showed better performance and a faster process, which could reduce barriers when developing high-performance CV products in pig production industry.

  PublisherDOI

• 307 Hothog: a Smartphone Application to Support Environmental Management Decisions for Non-Pregnant and Gestating Sows

  Journal of Animal Science · 2023-11-06

  articleOpen access

  Abstract Several management practices and technologies have been developed to mitigate thermal stress in swine. However, recommended temperature thresholds for implementing thermal stress mitigation are variable and may not accurately reflect the thermal requirements of non-pregnant and gestating sows with modern genetics. Therefore, a tool to support environmental management decisions for non-pregnant and gestating sows was created. The decision support tool is based on several published works by our group that characterized responses of non-pregnant, mid-gestation, and late-gestation sows across a wide temperature range. Cool, comfortable, and warm ranges were identified based upon behavioral thermal preferendum, and physiological responses were used to determine mild, moderate, and severe heat stress thresholds. The inflection points of respiration rate and body temperature as a function of dry bulb temperature were used to establish heat stress thresholds by gestation stage. Non-pregnant, mid-gestation, and late-gestation sow mild heat stress thresholds differed (P < 0.05) and occurred at 25.5, 25.1, and 24.0 °C, respectively. Body temperature inflection points indicative of moderate heat stress differed by gestation stage (P < 0.05) and occurred at 28.1, 27.8, and 25.5 °C, respectively. Severe heat stress inflection points were lower (P < 0.05) for late-gestation sows (30.8 °C) but were similar for non-pregnant and mid-gestation sows (32.9 °C). Dewpoint influenced heat stress response for mid- and late-gestation sows (P < 0.05) but did not have an effect on non-pregnant sows (P > 0.05). Heat stress threshold data were integrated with behavioral thermal preferendum data. For non-pregnant and mid-gestation sows, similar (P > 0.05) cool (< 13.2°C) and comfortable (13.2 to 16.4°C) temperature ranges were observed. However, late-gestation sows had lower (P < 0.05) cool (< 12.6°C) and comfortable (12.6 to 15.6°C) temperature ranges relative to non-pregnant and mid-gestation sows. The dry bulb temperature (TDB) that non-pregnant, mid-gestation, and late-gestation sows found to be warm was estimated as TDB preference range < TDB < mild heat stress. The decision support tool was integrated into a smartphone application called HotHog. This smartphone application provides hourly and daily predictions of thermal comfort and stress in non-pregnant, mid-gestation, and late-gestation sows. Users can set geographical locations, either manually or by current location, for thermal index predictions. Additionally, hourly and daily precipitation, and temperature predictions are displayed for the selected location. Management observations and mitigation options, as well as expected physiological and behavioral changes, are provided for each thermal index category to help users identify stressed pigs and assist with management decisions. HotHog will support swine producers in making more informed decisions related to in-barn environmental management to reduce the negative effects of thermal stress on sows and their gestating offspring. Furthermore, the tool may be helpful for teaching thermal management to livestock technicians and students.

  PublisherOA PDFDOI

• Evaluating the temperature preferences of sexually mature Duroc, Landrace, and Yorkshire boars

  Translational Animal Science · 2023-01-01 · 2 citations

  articleOpen access

  Abstract An accurate understanding of boar temperature preferences may allow the swine industry to design and utilize environmental control systems in boar facilities more precisely. Therefore, the study objective was to determine the temperature preferences of sexually mature Duroc, Landrace, and Yorkshire boars. Eighteen, 8.57 ± 0.10-mo-old boars (N = 6 Duroc, 6 Landrace, and 6 Yorkshire; 186.25 ± 2.25 kg) were individually tested in thermal apparatuses (12.20 m × 1.52 m × 1.86 m) that allowed free choice of their preferred temperature within a 8.92 to 27.92 ºC range. For analyses, the apparatuses were divided into five thermal zones (3.71 m2/thermal zone) with temperature recorded 1.17 m above the floor in the middle of each zone. Target temperatures for thermal zones 1 to 5 were 10, 15, 20, 25, and 30 ºC, respectively. All boars were given a 24-h acclimation phase followed by a 24-h testing phase within the thermal apparatuses. Daily feed allotments (3.63 kg/d) were provided to each boar and all boars were allowed to consume all feed prior to entering the thermal apparatus. Water was provided ad libitum within the thermal apparatuses with 1 waterer per thermal zone. During testing, boars were video recorded continuously to evaluate behavior (inactive, active, or other), posture (lying, standing, or other), and thermal zone the boar occupied. All parameters were recorded in 15 min intervals using instantaneous scan sampling. Data were analyzed using GLM in JMP 15. For the analyses, only time spent lying or inactive were used because they were observed most frequently (lying 80.02%, inactive 77.64%) and were deemed to be associated with comfort based on previous research. Percent time spent active (19.73%) or standing (15.87%) were associated with latrine or drinking activity and were too low to accurately analyze as an indicator of thermal preference. Breed did not affect temperature preference (P > 0.05). A cubic regression model determined that boars spent the majority of their time inactive at 25.50 ºC (P < 0.01) and lying (both sternal and lateral) at 25.90 ºC (P < 0.01). These data suggest that boar thermal preferences did not differ by breed and that boars prefer temperatures at the upper end of current guidelines (10.00 to 25.00 ºC).

  PublisherOA PDFDOI

• New Benchmark for Development and Evaluation of Computer Vision Models in Pig Farming Scenarios

  SSRN Electronic Journal · 2023-01-01 · 1 citations

  preprintOpen accessSenior author
  PublisherDOI

• Barriers to computer vision applications in pig production facilities

  Computers and Electronics in Agriculture · 2022 · 38 citations

  • Computer Science
  • Risk analysis (engineering)
  • Computer Science
  PublisherDOI

• A behavior and physiology-based decision support tool to predict thermal comfort and stress in non-pregnant, mid-gestation, and late-gestation sows

  Journal of Animal Science and Biotechnology/Journal of animal science and biotechnology · 2022-12-10 · 12 citations

  articleOpen access

  Abstract Background Although thermal indices have been proposed for swine, none to our knowledge differentiate by reproductive stage or predict thermal comfort using behavioral and physiological data. The study objective was to develop a behavior and physiology-based decision support tool to predict thermal comfort and stress in multiparous (3.28 ± 0.81) non-pregnant ( n = 11), mid-gestation ( n = 13), and late-gestation ( n = 12) sows. Results Regression analyses were performed using PROC MIXED in SAS 9.4 to determine the optimal environmental indicator [dry bulb temperature (T DB ) and dew point] of heat stress (HS) in non-pregnant, mid-gestation, and late-gestation sows with respiration rate (RR) and body temperature (T B ) successively used as the dependent variable in a cubic function. A linear relationship was observed for skin temperature (T S ) indicating that T DB rather than the sow HS response impacted T S and so T S was excluded from further analyses. Reproductive stage was significant for all analyses ( P < 0.05). Heat stress thresholds for each reproductive stage were calculated using the inflections points of RR for mild HS and T B for moderate and severe HS. Mild HS inflection points differed for non-pregnant, mid-gestation, and late gestation sows and occurred at 25.5, 25.1, and 24.0 °C, respectively. Moderate HS inflection points differed for non-pregnant, mid-gestation, and late gestation sows and occurred at 28.1, 27.8, and 25.5 °C, respectively. Severe HS inflection points were similar for non-pregnant and mid-gestation sows (32.9 °C) but differed for late-gestation sows (30.8 °C). These data were integrated with previously collected behavioral thermal preference data to estimate the T DB that non-pregnant, mid-gestation, and late-gestation sows found to be cool (T DB < T DB preference range), comfortable (T DB = T DB preference range), and warm (T DB preference range < T DB < mild HS). Conclusions The results of this study provide valuable information about thermal comfort and thermal stress thresholds in sows at three reproductive stages. The development of a behavior and physiology-based decision support tool to predict thermal comfort and stress in non-pregnant, mid-gestation, and late-gestation sows is expected to provide swine producers with a more accurate means of managing sow environments.

  PublisherOA PDFDOI

Frequent coauthors

• Richard S. Gates

  Iowa State University

  28 shared

• Daniel W Shike

  University of Illinois Urbana-Champaign

  17 shared

• Jay S Johnson

  Agricultural Research Service

  12 shared

• Brianna N. Gaskill

  Novartis (United States)

  11 shared

• Jiangong Li

  10 shared

• Ryan N. Dilger

  University of Illinois Urbana-Champaign

  10 shared

• Betty R. McConn

  Oak Ridge Institute for Science and Education

  9 shared

• Brett C. Ramirez

  Iowa State University

  8 shared

Education

• Bachelor of Science, Biosystems and Agricultural Engineering

  University of Kentucky

• Master of Science, Biosystems and Agricultural Engineering

  University of Kentucky

• Doctor of Philosophy, Agricultural and Biosystems Engineering

  Iowa State University