AI-Powered Tool Revolutionizes Detection of Ghost Fishing Nets

Iris Coleman
Jun 08, 2025 02:33
A new AI tool has been developed to identify and locate abandoned fishing nets, known as ghost nets, using advanced underwater imaging technology.
In a significant stride for marine conservation, a new AI tool has been developed to identify and locate abandoned fishing nets, commonly referred to as ghost nets. This technology aims to mitigate the detrimental effects these nets have on marine wildlife. According to NVIDIA’s blog, conservationists have introduced a platform capable of analyzing vast amounts of underwater imaging data to pinpoint these ghost nets.
The Environmental Impact of Ghost Nets
Every year, approximately 2% of global fishing gear, including about 80,000 square kilometers of fishing nets, is lost in the oceans. These nets pose a severe threat to marine wildlife such as seals, turtles, and dolphins, often leading to their entanglement and subsequent drowning. Additionally, as these nets decompose, they contribute significantly to plastic pollution in marine environments.
Technological Innovation in Detection
The detection of these ghost nets, which are often nearly invisible and as thin as a finger, presents a substantial challenge. To address this, WWF Germany, in collaboration with Accenture and Microsoft’s AI for Good Lab, has developed GhostNetZero.ai. This online platform aggregates high-resolution underwater data from various sources, including research institutes and offshore wind-power companies, to aid in the identification of ghost nets.
The AI tool utilizes a Convolutional Neural Network (CNN) with DeepLabV3 and ResNet50 backbone to scan sonar data for signs of ghost nets, achieving an impressive 94% accuracy rate. This precision allows for the pinpointing of ghost net locations, as each sonar image includes geolocation metadata.
Collaborative Efforts for Net Removal
Once a ghost net is identified, local divers and fishermen are engaged to confirm its presence and gather more detailed information, such as the size of the nets. These insights are critical for organizing the logistics required for net removal. WWF Germany’s project manager, Gabriele Dederer, emphasizes the importance of regional collaboration, stating that local protocols are essential due to the logistical and financial complexities involved in retrieving these nets.
The initiative is currently active in countries like France, Estonia, and Sweden, with plans for further expansion. The integration of AI into this project has been facilitated by Microsoft’s cloud services and NVIDIA’s GPU technology, enhancing the model’s accuracy and efficiency.
Future Prospects and Technological Advancements
Christian Bucher, a Microsoft representative involved in the project, highlighted the role of AI in pattern recognition, crucial for distinguishing the varied appearances of ghost nets in sonar images. The project continues to evolve, with local teams’ verifications feeding back into the AI model to improve its performance.
This innovative approach to addressing marine pollution represents a promising step forward in the global effort to protect marine ecosystems. By leveraging cutting-edge technology, the initiative not only identifies threats but also paves the way for effective removal and prevention strategies.
Image source: Shutterstock