Maritime Security - Illegal Fishing Detection Using Deep Learning

Illegal fishing is an unresolved and internationally pervasive problem that occurs both on the high seas and in areas within national jurisdiction. Existing technologies use anomaly detection algorithms that examine the data from Automatic Identification System (AIS) and radar cameras having long range. It is capable of covering 44.4 km. These algorithms also include clustering, density-based methods, or neural networks that are used to identify unusual behavior or patterns in fishing vessel data. These algorithms can flag vessels that deviate from typical fishing patterns, helping authorities target suspicious activities. However, they fail to identify the complex patterns and accurately detect illegal fishing. We propose a system that increases the accuracy of algorithms by identifying complex patterns to send an alert to respective authorities. These alerts play a role in decreasing illegal fishing to some extent. Based on [1] US Food and Agriculture Organization’s illegal fishing activities have an economic value of US${\$}$10- 23 billion (estimates) and are responsible for the loss of 11-26 million tons of fish every year. We are proposing a system based on remote sensing data i.e., satellite images with higher resolution