Excellent precision, saved time and increased volumes with AI and Machine Learning
The university of Gothenburg and Nationalpark of Kosterhavet needed a better way of understanding the situation in the sea and how to help the environment in the best possible way. Meassurements were exteremely difficult to make under water, with moving objects, several spicies and massive amounts of images and data to go through manualy. They contacted Combine and AI and Machine Learning was used by Combine to process images, create algorithms and present tangible results in form of object data.
This expertize and the same methods are applicable and beneficial for many other areas as well, for example autonomous driving, medical scanning, traffic planning and so on.
Challenge: To perform regular investigations of the seas and to compare the environment, the distribution of species and the development over time takes enormous amount of manual work, time and money. With the old solution, it was not possible to monitor the seas in a satisfying way, and therefore not possible to really understand how to help the environment in a specific place.
Objective: Create an intelligent end-2-end solution where data in form of underwaterpictures is put in and data of the specific area comes out. The solution should be self-learning and continue to improve with every piece of data that is put in.
Solution: We used a machine-learning system and built algorithms that detects certain objects in all the pictures that was fed into the system. By using the algorithm, the machine learned how to separate and count certain spicies and present a direct result, the algorithm then improves itself as a result of the calculations and results.
Result: It is now possible to make more frequent investigations, compare data year-over-year, get results much quicker and free up a lot of time for the scientists to make conclusions and improvements instead of gathering data. This is a massive improvement compared to of years of manual studies.
Key to success: Combine has thorough expertize in the area of image processing and building algorithms. These experts worked togehter with the scientists and end-users in a tight team to secure that the solutions was not only good in a technical perspective, but also exactly what the users needed to create value for them.