We’re excited to report on a new, cost-effective, few-shot classification AI model to help solve environmental challenges!

e3Scientific has teamed up with the amazing team at Intranel to develop a cost-effective few-shot classification methodology to identify historic sheep dip locations.

Using AI foundation model features (DINOv2) and multi-modal large language model (GPT-4o) prompt engineering we can efficiently locate probable sites using only a small set of reference data – without finetuning underlying models.

An initial trial for Environment Southland used 186 aerial images and found over 300 potential historic sheep dip sites suitable for further investigation.

We are now improving the initial sheep dip model and exploring other applications of the model including identifying and monitoring wetlands, weeds and other vegetation.

Contact us if you would like more information or have an environmental problem that you think could be solved using the power of AI.

#AI #GenerativeAI #SheepDips

(AI-generated illustration of “The Problem With Sheep Dipping”. Created using DALL-E through OpenAI’s ChatGPT.)