AgrInfo needed a practical way to digitize pig-farm management tasks such as barn monitoring, animal counting, and disease screening because manual oversight was slow, costly, and difficult to scale across large livestock operations.
Reported impact
AgrInfo reduced manual farm-inspection effort, improved herd visibility and disease screening, and showed that livestock know-how can be packaged into reusable AI services for day-to-day farm management.
AgrInfo used Baidu's EasyDL customization and serving platform to turn everyday pig-farm operations into computer-vision workflows that ordinary field teams could use. The company trained three models in about a week: an electronic-fence detector that monitors barns and identifies people or pigs entering key areas, an AI counting model that estimates pig numbers from ceiling-mounted camera images, and a pig-disease recognition model that analyzes body images for common symptoms before escalating atypical cases to remote experts. These models were embedded into applications such as Remote-eye and Pig Disease Assistant so users could access them through cameras, mobile uploads, and mini-program workflows rather than data-science tooling. The broader value is that EasyDL let AgrInfo encode husbandry know-how into reusable AI services, connecting video frames, object detection, disease screening, and remote diagnostics in one agricultural management pipeline.
Barn-camera images and farmer-captured pig photos enter EasyDL-trained detection models, which classify presence, count animals, and screen for visible disease traits. The results are sent into Remote-eye and Pig Disease Assistant workflows, where users monitor farms remotely, review suspected health issues, and escalate non-typical cases to online experts.
AgrInfo trained electronic-fence, pig-counting, and pig-disease models with EasyDL in just one week (original: 仅在一周内便训练出"电子围栏""猪只盘点""猪病识别"三个模型).
Remote-eye enables remote monitoring and pig counting, reducing manual farm visits (original: 在"Remote-eye"猪只检测的帮助下，养殖户们可以实现远程监控猪场情况，实时盘点猪只数目).
The disease-recognition workflow can help detect conditions such as foot-and-mouth disease and skin disease from mobile photo uploads (original: "猪病通"猪病识别就能协助养殖人员判断猪只是否有口蹄疫、皮肤病等典型猪病).
The electronic fence uses EasyDL object detection to monitor whether people or pigs appear in monitored areas (original: 基于百度大脑EasyDL的物体检测能力，它可以实时检测区域内是否有人、猪的出现).
The models all achieved around 80% accuracy (original: 准确率均达80%左右).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Google Cloud Sample Modified · Oct 9, 2026 Luxia Agro Luxia Agro is an Argentine company specializing in the import of phytosanitary products for the country's agricultural sector. With a core team of just 10 people, the company has built a reputation for quality and… 50% reduction in labor hours Cloud Run · Cloud SQL · Cloud Storage 
NVIDIA Sample Sep 30, 2026 Corteva Corteva needed to evaluate billions of host-pathogen protein interactions and massive metabolomics datasets faster, but early-stage agricultural discovery pipelines were too computationally expensive to prioritize… 15-fold BioNeMo Boltz-2 NIM · NVIDIA H200 GPUs · NVIDIA cuML NVIDIA Sample Sep 4, 2026 Aigen Aigen needed to train field-robot perception and control systems for highly variable agricultural environments, but collecting and labeling enough real-world crop and weed data was too slow, seasonal, and expensive to… 99% synthetic data and 1% real-world data Alchemy · Cosmos Transfer 2.5 · NVIDIA Cosmos 3 
