As a Computer Vision Engineer at Clean Sea Solutions, I worked on developing the vision system for their autonomous ocean cleanup drone. The goal was to enable real-time marine debris detection using a fast and efficient computer vision pipeline.
My work focused on:
• Experimenting with and benchmarking various computer vision models for trash detection
• Training and evaluating models tailored to the challenges of the marine environment
• Prioritizing real-time inference and optimizing lightweight models for deployment
• Supporting integration with the drone's navigation and control systems
The project required balancing theoretical understanding with practical implementation, ensuring that the system could perform reliably in dynamic and unpredictable conditions at sea.
This experience gave me valuable insights into building production-ready computer vision systems capable of robust, real-time performance in the real world.