Embedded Files
Summary
Summary
The Ocean Cleanup is using artificial intelligence algorithms to identify and remove plastics from natural environments before they harm wildlife.
Contributor
Contributor
The Ocean Cleanup
The Conservation Challenge
The Conservation Challenge
Marine litter, particularly plastics, have become ubiquitous in the ocean. Large oceanic scale accumulation of buoyant plastic have been reported in the subtropical waters of the five main oceans. The persistence of some plastics at the surface of the ocean increases the risk of ingestion and entanglement for marine species. Plastics can also carry invasive species posing a risk on biodiversity and ecosystems. A better understanding of the source, the distribution, and the degradation of plastic in oceans are required to quantify the risks associated with plastic pollution.
The Solution
The Solution
To understand the distribution of marine debris at the surface of the ocean, The Ocean Cleanup conducts experiments for remote sensing of plastic by deploying Unmanned Autonomous Vehicles at sea. These vehicles are equipped with several sensors including RGB and multispectral cameras. Plastics have specific signature in the shortwave infrared wavelengths and spectral imagery are used to differentiate plastic from other objects and sun glints.
Description of Technology Used
Description of Technology Used
The Ocean Cleanup is using a neural network approach to identify the plastics based on RGB imagery. Based on public libraries and datasets labelled by volunteers, The Ocean Cleanup uses Google's TensorFlow to apply a combination of neural network layers to the imagery. The public dataset, named COCO, helps the model to detect objects from their background but has no information on marine litter. To include the marine litter information into the model, The Ocean Cleanup is partnering with Zooniverse, the largest online citizen science platform in the world. The images of plastics classified by citizen scientists will enable The Ocean Cleanup to refine the accuracy of the models.
jakarta_video_testrun.mp4Lessons learned
Lessons learned
To efficiently detect plastics from imagery, the quality of the image is crucial. Sun glints, sea foam, different camera exposures, and the size of the plastics influence in the accuracy of the AI algorithms. The Ocean Cleanup is also working to reduce the power consumption required by the AI algorithms. Currently, the Neural Network algorithms demand a significant amount of electricity and every bit of power is vital when you are in the middle of the ocean. In addition to these difficulties associated with the artificial intelligence system, The Ocean Cleanup is constantly discovering new ways to overcome engineering and scientific challenges.
Opportunities/Call To Action
Opportunities/Call To Action
You can financially support the project so that The Ocean Cleanup team can continue developing advanced technologies to rid the world’s oceans of plastic. If you want to play a key role in launching the largest cleanup in history, check out The Ocean Cleanup's website for job openings. Additionally, the The Ocean Cleanup is always open to collaborate with other scientists and engineers so feel free to get in touch with them.
Next Steps
Next Steps
The Ocean Cleanup follows an iterative design process, which entails testing and learning until they have a proven concept. Research plays an important role in monitoring the cleanup progress. Understanding where ocean plastic is coming from, where it is accumulating, and what happens to it in the long term are essential. To better understand these processes, the research team is investigating innovative methods to collect more field data. Check their latest updates to learn more about their work.
Report abuse