The project involves developing a pipeline inspection system using drones. The objective is to create a system that can identify potential oil leaks and detect possible problems in the pipelines, such as corrosion, wear and tear, and others.

The project began its development in 2021, where we trained a neural network to identify oil leaks in images of leaks we found on the Internet. However, we faced a major issue as we couldn't obtain images of oil pipelines with leaks to balance the dataset. Therefore, we decided to temporarily halt the project.

However, in 2022, we had the idea to build our own database. For this, we used our own mobile phones to capture photos of water pipelines with and without leaks. After capturing the images, we used software to identify the pipelines and then another software to identify the leaks. With this, we were able to build a database with more than 4000 images of water pipelines with and without leaks.

The results of our project were promising, having been published in the annual congress of the Brazilian telecommunications society, and now we are working to improve the accuracy of our system and expand its capabilities. We believe that our solution has the potential to      revolutionize the way oil pipelines are inspected, and we are excited to continue working on this project in the future. We plan to include temperature-based images of oil pipelines.