Using Computer Vision to Detect Dangerous Sea Immigration

Together with impact-driven startup ACUA Ocean 50 AI changemakers developed a cost-effective solution to detect and potentially stop dangerous sea immigration.

The problem 

Last year there were 95,000 illegal sea crossings from Africa into Europe. The criminal smuggling of migrants by sea is one of the most dangerous forms of migrant smuggling and often requires urgent humanitarian assistance. Since 2015 over 18,000 people have died or are believed missing in the Mediterranean and Atlantic in their attempt to reach Europe, while an additional 541,600 people have needed to be rescued in the Mediterranean alone. To save the lives of those in distress at sea, EU countries’ coastguards and navies have deployed substantial resources, however, due to the nature of marine operations, the cost of doing so is high and therefore limited. Between 2006 to 2017, the EU spent at least €676.4 million on maritime “walls”.

ACUA Ocean is developing a cost-effective Autonomous Surface Vessel (H-USV) solution that is aimed at reducing the costs of these efforts whilst providing a greater presence at sea to save more lives.

The project outcomes

The team went through the entire data science project lifecycle, starting with data collection, labeling, and testing different computer vision approaches. Within eight weeks, the team built a computer vision model that’s compatible with FLIR camera systems mounted on boats to identify vessels and individuals potentially conducting illegal and life-threatening sea immigration operations.