Project
Computer Vision & IoT for Precision Agriculture
Low-cost vision and sensing for animal welfare — tracking broiler-chicken behaviors and managing dairy-herd disease with USDA-funded UT collaborations.
Animal agriculture generates enormous welfare and health data that nobody can watch — tens of thousands of broiler chickens per house, herds whose diseases spread faster than symptoms appear. Two USDA-funded UT collaborations apply AICIP’s vision and machine learning expertise to that gap.
The poultry welfare project (USDA NIFA AFRI, $1M) develops lightweight deep learning to track welfare-indicating behaviors of individual broilers — stretching, preening, dustbathing, feeding — from low-cost cameras, targeting a system economical enough (~$2,500 per house) for commercial adoption. The team is led by the UT Institute of Agriculture with AICIP contributing the computer-vision methods.
The dairy herd disease management project (USDA NIFA, $1M) pairs a biosensor with GPS ear tags, temperature sensing, and an IoT network at UTIA’s Little River research dairy, feeding a digital-twin model of disease spread. AICIP’s role centers on pattern recognition that keeps device readings accurate on noisy field data.
Both efforts reflect a lab habit: methods built for hard perception problems travel well into domains where sensing is cheap and labels are scarce.