How do you help train AI, protect Kakadu and win a prize? Now that’s innovation
Image caption: Types of labelled imagery and use cases
Written by Heritage, Reef and Wildlife division
Calling on citizen scientists to help ID fish and protect the park
Kakadu National Park has remained protected from the effects of uranium mining for more than 40 years.
Technological advancements and safety driven innovation means we can keep better track of freshwater fish communities. This allows us to detect mining related impacts on the park. Scientists now use drones and underwater video cameras. This helps to safely conduct annual surveys of billabongs throughout the national park. The cameras allow our scientists to measure the presence and abundance of freshwater fish species. It also enables the collection of hundreds of hours of fish videography every year.
We are now working with Microsoft engineers to develop an innovative artificial intelligence (AI) model. It will help automate the identification of fish species.
When completed, the AI model will be able to identify and count fish from video. This will significantly increase the efficiency and accuracy in our data processing. But we need your help and are calling on the support of citizen scientists!
In order to fully automate the process, the model needs to be trained. This is done by providing more than 100,000 annotated and labelled images of fish. If you can help with the annotations, your actions will help train artificial intelligence. There is also a prize for the participant with the highest number of labelled images!