Intelligence that understands the field.

Technology that goes beyond the obvious.

Multimodal artificial intelligence, computer vision and agronomic knowledge to investigate problems on your farm.

Multimodal intelligence

A farm can tell its story in many ways.

The Cocoa App combines different types of information to better understand the reported problem.

Voice

The farmer describes symptoms by audio. The app understands the context and the language used in the field.

Text

Questions and information can be sent in a simple way.

Image

Photos of leaves, pods and stems can be used to deepen the investigation.

Geographic intelligence

Recommendations adjusted to the climate and soil of the region can support an analysis closer to the reality of the farm.

Computer vision

The image is also part of the investigation.

The Cocoa App uses visual analysis to identify patterns related to pests, diseases and nutrient deficiencies.

Images can be sent in the Diagnosis Room when a deeper investigation is needed.

Photo straight from the field
Analysis with artificial intelligence
More precise diagnosis
Folha de cacaueiro com sintomas analisados pela visão computacional
Leaf spot Likely fungal disease
Fruto de cacau no campo com sintomas de podridão-parda
  • Plant pathology
  • Plant nutrition
  • Soils
  • Field experience
Agronomic knowledge

Technology built from knowledge about cocoa.

The Cocoa App was developed with contributions from specialists in different areas related to cocoa.

This knowledge is used to build and train the artificial intelligence.

The model today

A first step with a large impact.

In this first launch phase, the Cocoa App supports visual and symptom-based identification of a set of common problems in cocoa farms in southern Bahia.

8diseases Main diseases affecting cocoa in the region.
11pests The most frequent insects and pests in the fields.
+nutrient deficiencies Changes related to management and soil.
Folha de cacaueiro com sintomas de vassoura-de-bruxa
Detalhe de folha com manchas identificadas pelo App do Cacau Witches’ broom spot Preliminary identification
Next steps

More knowledge for a stronger farm.

The Cocoa App is constantly evolving. In the next phases, the project aims to expand the number of identified problems and include new management features.

More diseases, pests and deficiencies

Expanding the number of problems covered by the algorithm.

New management features

Adding recommendations on fertilization, pruning and other practices.

Wageningen model

Integration of the visual identification model developed by Wageningen University, in the Netherlands.

Continuous evolution

New data, images and technical knowledge to make the Cocoa App increasingly complete and useful for farmers.