Savimbo is engaging KUNGFU.AI to advise and support the development of a classifier for forest imagery sourced from the Colombian Amazon to verify the current state of the forest for specific locations. The technical infrastructure development, all data manipulation, and model training/tuning will be performed by a Savimbo partner. KUNGFU.AI will support these efforts as appropriate and advise on both the general approach and technical implementation.
Savimbo's challenge was fairly straightforward. They had a set of ground level landscape images from the Columbian rain forest and they wanted to build a basic classification model. The images were labeled in groups such as 1-year growth vs. 5-year growth, and the model simply needed to be trained to classify new images.
Their ultimate goal was to use this as part of a fundraising campaign to create a marketplace for carbon credits. Competing solutions relied on satellite imagery, and the differentiation here is that they worked directly with farmers to compensate them for maintaining rain forest habitat.
Savimbo hired a third party development team from Mexico to build the pipeline and train the computer vision classification model. They engaged KUNGFU.AI to act as a technical advisor to guide the work. KUNGFU.AI was not responsible for coding, but we did support their deployment.
Although it's still early to tell, this work has potential to be groundbreaking. The work we did with them is a key component of their efforts and a core part of their ability to enforce the carbon credit marketplace they’re trying to create. Savimbo also mentioned the possibility of applying this to another rain forest habitat and they assumed it would require a separate model due to differences in local vegetation.
Savimbo is engaging KUNGFU.AI to advise and support the development of a classifier for forest imagery sourced from the Colombian Amazon to verify the current state of the forest for specific locations. The technical infrastructure development, all data manipulation, and model training/tuning will be performed by a Savimbo partner. KUNGFU.AI will support these efforts as appropriate and advise on both the general approach and technical implementation.
Savimbo's challenge was fairly straightforward. They had a set of ground level landscape images from the Columbian rain forest and they wanted to build a basic classification model. The images were labeled in groups such as 1-year growth vs. 5-year growth, and the model simply needed to be trained to classify new images.
Their ultimate goal was to use this as part of a fundraising campaign to create a marketplace for carbon credits. Competing solutions relied on satellite imagery, and the differentiation here is that they worked directly with farmers to compensate them for maintaining rain forest habitat.
Savimbo hired a third party development team from Mexico to build the pipeline and train the computer vision classification model. They engaged KUNGFU.AI to act as a technical advisor to guide the work. KUNGFU.AI was not responsible for coding, but we did support their deployment.
Although it's still early to tell, this work has potential to be groundbreaking. The work we did with them is a key component of their efforts and a core part of their ability to enforce the carbon credit marketplace they’re trying to create. Savimbo also mentioned the possibility of applying this to another rain forest habitat and they assumed it would require a separate model due to differences in local vegetation.
