Research Line - Smart Harvest

Make sense in agriculture of the future!

Research themes

Increased efficiency

  • We work to improve the efficiency of mechanized agricultural process.

Improve quality

  • Our research seeks to obtain the best quality of the harvested products.

Decrease losses

  • We believe it is possible to reduce food waste.

PhD thesis

Deep learning image processing to estimate peanut harvest losses

Armando Lopes Brito Filho

Reducing losses, increasing food safety!

The management of the mechanized harvesting operation is essential for controlling production and providing economic returns for agricultural activities.
In this process, loss detection is the key to monitoring the operational quality of the mechanized systems. Thus, our research aims to develop new technologies using automated image processing to detect, count, and estimate losses for mechanized peanut harvesting.

PhD thesis
Recycling by-products from harvesting into antioxidants to solid biofuels

 Bruno Rafael de Almeida Moreira

Power the world without threat!

Our focus is to elaborate sustainable waste-to-product paths. Thereby, we will analyze the feasibility of recycling vegetable losses from harvesting coffee, cotton, corn, peanut and sugarcane into anti-oxidative additives to prevent fuel-flexible pellets from emitting airborne fast-acting toxic chemicals indoors. Our insights will provide forward knowledge of particular relevance to progress in the field's prominence in operating safety-sensitive logistics for solid biofuels towards occupational health and industrial hygiene. They also will act as an opening of eco-compatible solutions to broaden and strengthen the emerging yet exciting smart harvest network. Overall, our study will serve as a springboard to accelerate the efforts to mitigate global challenges ahead.

PhD thesis

Operational and energy performance of tractors as a function of the constructive types of agricultural tires

Edward Victor Aleixo 

We respect and protect the soil!

Tractors used in agricultural operations have undergone changes, which include increasing horsepower and using different traction devices such as diagonal and radial tires and rubber tracks.

Recently, an agricultural tire with rubber track characteristics was made available to the market for use in agricultural tractors, with the objective of providing less soil compaction, higher traction coefficient, operator comfort and lateral stability to the tractor for work on rough terrain.

Our research, pioneering in Latin America, aims to evaluate the operational and energy performance of agricultural tractors, when using diagonal tires and track tires in different operations.

PhD thesis

Machine learning for on-farm cotton yield prediction based on satellite images

Francielle Morelli Ferreira

Technology diffusion using Artificial Intelligence in agriculture!

The big data scenario and the exploration of artificial intelligence in agriculture allow studies on the various information obtained in the field by the machines, combined with the use of data available for free on digital platforms how time series of satellite images, climate parameters, harvest history, etc.

The adoption of these tools for generating results is becoming more popular, enabling the use of machine learning algorithms in agricultural data being directed to digital agriculture, acting in a faster, more efficient, and more accurate way.

We are conducting research aiming technology diffusion, to predict information, such as yield, obtaining great results, enabling its use as a technique that will help farmers and data managers, resulting in time savings, whether in planning, marketing, or intervention in harvest management. 

PhD thesis 

Peanut yield estimation using remote sensing and artificial intelligence tools

Jarlyson Brunno Costa Souza


The sky is not the limit, it's the solution!

The sustainable development in agriculture involves the application of new technologies that aim to increase production to reduce environmental impacts. The use of remote sensing and artificial intelligence tools can increase productivity in agricultural fields, reducing losses and helping the producer in decision making. For peanuts, the estimation of agronomic parameters of the crop can bring great development to the sector. The determination of the maturation index and productivity remotely can improve the use of resources and bring better management for the producer, increasing profitability and quality of the harvested products.

PhD thesis

High-Throughput Analysis in Sugarcane Based on UAV imagery

Marcelo Rodrigues Barbosa Júnior 

Toward harvesting high-quality material!

Problems associated with the increased efficiency and improved quality and pricing of the harvested product call for the use of high-accuracy approaches and thus have received close attention from researchers and practitioners. Our research uses a high-resolution remote sensing platform applied in high-throughput analysis in sugarcane to show that:

i) estimating sugar content help to determine the best time to harvest;

ii) harvest management is a potential fit-to-purpose strategy can marginally help to avoid

Masters dissertation



Artificial intelligence: Taking the Agriculture Industry into the future!

The introduction of technologies in agricultural activities is necessary to achieve the economic environmental demands of the coming years. Inserting, machine learning methods and artificial intelligence, in agriculture are means of achieving the goals of productivity growth and reduction of environmental impacts. Thus, our project proposes to develop an artificial neural network, using machine learning, image processing data, vegetation indices, soil and crop characterization, to estimate sugar beet yield and quality.

Masters dissertation

A framework based on object detection and instrumentation to monitor peanut yield during mechanized harvesting
 Vinícius dos Santos Carreira

Forget the ordinary, risk your ideas!

Our team is developing a digital tool to monitor peanut yield during mechanized harvesting. The main idea is to characterize the spatial variability and allow adjustments in the machine settings. This framework can also be used for other below-ground crops.