Inspiration
Inspired by two guys which presented the deforestation challenge
What it does
Our solution makes the segmentation of the satellite images where it detects the forest ares.
How we built it
In core of our model there is SOTA semantic segmentation model Segment Everything (SAM). We fine-tuned our model using LoRA approach, which uses smaller model to optimize SAM for our specific task. It has been trained on over 2500 images.
Challenges we ran into
Understand sustainability market needs
Accomplishments that we're proud of
we developed awesome prototype of potential product which can help companies to comply with EUDR regulations and of course to solve environmental problems
What we learned
Knew a lot about sustainable markets, new approaches to develop working ai model, real world problems usage of our application
What's next for Lesok
sell forest to china :)
Log in or sign up for Devpost to join the conversation.