Edge AI at the farm: precise livestock management helps farmers optimize livestock productivity.
The livestock industry is huge, however, several issues impede its productivity, such as the soaring feed prices due to extreme weather conditions, disease risk, and environmental and pollution regulations. Now, AI is helping to modernize livestock farms to track and detect livestock diseases and weight.
Precise breeding is necessary to improve productivity, but it is difficult to achieve due to human workload constraints. . Intflow, a Korean company that focuses on computer vision solutions, aims to alleviate this problem with on-site cameras and Seeed’s reComputer edge AI devices.
Livestock management powered by Intflow’s AI and Seeed’s reComputer J1010
Intflow provides EdgeFarm, an AI solution that perceives livestock injuries and diseases to help farmers manage and optimize livestock productivity. EdgeFarm obtains the biometric data of each 40 piglets for each ceiling-mounted camera. It measures real-time data of the pigs, for example, their eating and exercising habits, how many times they fought with one another, and how much weight they gained in a day. Based on this data, the current breeding situation will be judged, and if it is not optimal, an action list will be provided to improve on the current situation. If the past livestock industry is compared to a large truck driven without a car dashboard, EdgeFarm acts as a digital dashboard for the driving.
In 2022, Intflow is expanding its edge farm business to manage 20,000 pigs 24 hours a day, and for this purpose, a commercialized Jetson Nano solution is essential. Seeed Studio helps EdgeFarm’s global expansion with the hardware and fulfillment. The system also supports cattle management.
“With Seeed’s reComputer J1010, we can reduce the management cost per animal by 98% compared to the competing solution that relies on GPU-cloud because the Edge AI solution with Jetson could provide the lowest inference cost per a camera channel,” said Kwang Myung Jeon, CEO at Intflow Inc.
Furthermore, reComputer J1010 allows more possibility of IoT at the edge combining AI with image analysis. With IoT sensors plugged into reComputer, farmers can monitor the humidity and environmental temperature, animal and farm status, track the information, and control the production safety during the growth process. Periodically, the system tracks key information concerning disease prevention and breeding progress.
EdgeFarm uses low-cost IP cameras to monitor each sector of the farm. Each camera is connected to the dedicated edge AI device that conducts AI analysis on-site. Currently, they are using reComputer J1010 to empower up to eight IP-camera with AI inference capabilities.
Even for the latest computer vision technology, tracking hundreds of pigs in real-time is a burdensome task Therefore, we designed object detection, ID tracking, and action recognition suitable for animals in a highly optimized way. Specifically, we adopted a rotated bounding box with key points for exact detection even in densely packed scenarios which are very common in livestock farms. The proprietary AI is optimized through NVIDIA’s TensorRT and then loaded onto the Jetson Nano to achieve the performance of processing up to eight cameras simultaneously. The biometric information of each camera obtained from the Jetson Nano passes through the EdgeFarm Database into the cloud and delivers the information to the web dashboard and app.
The Edge Farm roadmap released in 2022 includes pig inventory management and growth management solutions. The inventory management solution measures the number of pigs passing in the hallway and their weight in real-time and it immediately creates a report and delivers it to the farmer upon completion of the work. Through this information, the farmer can determine the number of pigs currently in the farm and how many are sold in their respective condition.
In the pig growth management solution, each camera is dedicated to a certain area, tracking the animals for 24 hours and acquiring biometric information. Combining object detection, tracking, and behavior identification, AI measures the amount of exercise, weight, and food for each animal without attaching any device. Based on the information measured every day, it is possible to maximize the farm’s profitability by predicting the optimal shipment time for each room. In addition, emergency situations such as fighting or death are identified and made aware to workers to prevent the spread of infectious diseases on the farm.
Intflow is a deep-tech startup founded in 2019 with the goal of eliminating industrial inefficiencies by developing the world’s best non-contact biometric information analysis technology. We focus on developing and distributing AI technology specialized for industrial sites, centering on the CEO, who obtained a Ph.D. in computer science from Gwangju Institute of Science and Technology (GIST), Korea. By developing ‘EdgeFarm’, a livestock animal digital care solution using machine vision technology, Intflow is achieving more than doubling growth every year. Intflow’s technology categories include machine vision, auditory intelligence, AI lightweight, edge computing, and animal behavior. Intflow’s top mission is to develop innovative technologies to lead the unmanned livestock industry and to pursue a prosperous world through the coexistence of technology and mankind.
About reComputer J10 and J20 series
reComputer series for Jetson are compact edge computers built with NVIDIA advanced AI embedded systems: J10 (Nano 4GB) and J20 (Jetson Xavier NX 8GB and Jetson Xavier 16GB). With rich extension modules, industrial peripherals, and thermal management, reComputer for Jetson is ready to help you accelerate and scale the next-gen AI product by deploying popular DNN models and ML frameworks to the edge and inferencing with high performance.