CODING.WATERKANT
CAMP

Applying Software Engineering Best Practices to AI Projects. Most AI Project are in a Demo state, we want to expand our capabilities into long-term qualtity ai software

Marine scientists increasingly rely on underwater imagery to study and monitor the seafloor. However, creating reliable datasets for image matching remains a major challenge because underwater environments are highly dynamic, visually complex, and rarely provide accurate ground truth. Factors such as water turbidity, changing illumination, suspended particles, and repetitive textures make establishing correspondences between images difficult, limiting the development and evaluation of computer vision algorithms.

Our project aims to tackle this challenge by building a citizen science platform that leverages gamification to involve the public in scientific data annotation. Through an intuitive and engaging web interface, users are invited to identify matching features between pairs of underwater images. By turning annotation into an interactive experience, we enable large-scale collection of human-generated image correspondences while making marine science accessible to a broader audience.

The resulting annotations will be aggregated into a high-quality dataset of underwater image matches that can serve as valuable ground truth for research in computer vision, robotics, and autonomous underwater exploration. By combining human intelligence with modern data-driven methods, our platform helps bridge the gap between citizen participation and scientific discovery, ultimately contributing to more robust algorithms for mapping, monitoring, and exploring the ocean floor.

...

This project aims to simplify the reimbursement process for the EXIST start-up grant by automating expense classification and application preparation. Users can upload invoices, and the system automatically extracts relevant information, identifies the appropriate expense category and cost item, and transfers the validated data into a complete reimbursement application. The underlying solution is designed to be transferable to other funding programs and administrative workflows, making it adaptable to a wide range of document processing and expense management use cases.

Groundwater levels vary throughout the year and depend on weather conditions. Although groundwater levels are routinely measured, we aim to predict future groundwater levels using publicly available data, such as weather observations and current groundwater measurements. In addition to forecasting groundwater levels, the proposed model provides insight into the most influential factors affecting groundwater level dynamics