Let's build great things together
The Tech Stack:
Ruby on rails
The Jetruby FMS is a highly customisable AI & ML platform that will be expertly tailored to solve your specific needs.
EveryPig is a mobile solution for pork producers, delivering real-time advice based on AI and ML technologies. By processing thousands of relevant images, the application evaluates the health condition of pigs and signals about a possible outbreak of infection.
Among other solutions, the project provides the following:
Collecting real-time data for production and veterinary analysis;
Veterinary checkups in the digital format and telemedicine tools;
Management and communication solutions for pig farms.
Improving the accuracy of image recognition
Training Machine Learning mechanisms to evaluate images and return accurate information results has been a huge part of the entire process. Getting consistent and actionable results with voluminous data is one of the core puzzles in machine learning. Poor quality of photo images or poor lighting could affect algorithms of neural networks that are subject to a high degree of fluctuation.
Thanks to the successful implementation of machine learning practices, it became possible to efficiently analyze signs of diseases and identify up to 30 diagnoses by photos. Several thousands images were processed to achieve this ambitious result.
Often, the app is used in remote barns and farms. It was important to ensure its speedy operation without overlooking the aspect of caching. Otherwise, sending high-res images and receiving the information results would be problematic. Improved caching enabled a smooth rendering of images.
The application required essential improvements to transform the user experience into a smooth journey.
One of the major tasks was to develop the mobile interface, allowing farmers to send images of their pigs and obtain a likely diagnosis.
Following the project guidelines and specifications, we made the app faster and more user-friendly. It became possible to process images in bulk and receive diagnosis recommendations at a normal speed above 110.07 Mbps.
Making Data Serve the End-User
ML algorithms worked on the collection of photos featuring the affected organs of dead pigs. To process a huge volume of images, it was necessary to send them through the back-end. Next to it, the neural network started to classify the data according to certain characteristics.
Developers trained neural networks so they could see and detect regularities in provided photos. Algorithms analyzed signs of abnormalities featured on photos and worked to associate them with one of about 30 diagnoses. Finally, the app notified a farmer about a probable diagnosis.
By getting neural networks to recognize disease patterns, we improved the quality of recognition by up to 80% and, thereby diagnosis accuracy.
We’ve developed a fast, informative, and “one-stop” platform that has become a game-changer for pork producers. It inspired the business owner to continue the expansion of his enterprise across the globe.