BeefTrader: market information intelligence platform for maximizing the profit of producers in the meat industry

Abstract

Brazil is the world's largest exporter of meat, serving 157 countries, accounting for 18% of the international market and having the largest meatpacking and distribution networks on the planet. The beef cattle agribusiness movement represents the largest gross value sector of Brazilian agricultural production and generated R$ 747 billion in 2020. In the world, this complex, in 2016, moved US$ 315 trillion. Increasing the profitability of farmers can impact 10% of the Brazilian GDP. This project aims to stimulate the increasing of productivity, improve profitability and reduce the cost of @ produced, decommodifying it, helping countries meet their demands, with returns for producers, processors and consumers. The proposal is to offer a set of mechanisms to monitor animals from rearing and feedlot to slaughter, in addition to allowing monitoring of weight, BCS and frame size based on the characteristics of the individual and formation of groups of animals more profitable using computational vision, Artificial Intelligence (AI) algorithms integrated with operational research decision-making models. This proposal will bring better traceability, transparency, better use of natural resources, reducing the environmental impact and food insecurity of livestock farming, which will still earn more profit inside and outside the gate. In the development of BeefTrader (process FAPESP 2015/07855- 7) and BeefTrader Grass (FAPESP Project 2019/09134-6) in PIPE Phase 1 models, algorithms, publications and results of proof of concept were generated demonstrating their feasibility mainly within the farm frontier. In PIPE Phase 2, the main objective of the research will be to automate the collection of measurement and evaluation of animal characteristics for weight, BCS and frame size, coupling computer vision intelligence, which gives the small company less technological dependence on sensors from other companies, better cost-benefit ratio for the producer, thus helping to scale the solutions on the market, improving the profitability of the business. The steps to achieve these goals are: 1) development of a sensor for collecting measurements from animals (hardware and 3DBeef system), collecting input variables for the individual characterization algorithms for beef cattle; 2) improve 3DBeef algorithms: platform for individual characterization of beef cattle using computer vision and machine learning to estimate weight, BCS, frame size that will culminate in an AI algorithm (as an integrative innovation of the proposal) for the formation of homogeneous batches of more profitable animals; 3) improve BeefTrader and BeefTrader Grass with the integration of predictive growth, tissue deposition and profitability models fed back by the 3DBeef system. Up to 15,000 animals from farms already monitored or interested in partnership with @Tech will be monitored with BeefTrader, BeefTrader Grass and 3DBeef. On the farm, the animals will be: identified, weighed, characterized and automatically evaluated via BeefTrader, BeefTrader Grass and 3DBeef in vivo. (AU)