AI recognition of electric transmission towers

AI recognition of electric transmission towers

Video
Team name / Company name: Senzory do škol
Team leader: Alexandr Waage
Challenge: no. 9: Grid Control by drones – picture recognition
Problem: The problem simply deals with the modernization of the issue. At present, ČEPS conducts inspections every year and has done so manually. However, there was pressure to automate and accelerate, specifically recognizing the types of towers and recognizing errors and faults.
Solution: Basically, as we have the task of recognizing the specified data, so at the moment we already recognize the given types. More is in presentation here: https://docs.google.com/presentation/d/1pGla0t-YNanrxGAN7DYVcZjM0E5dYby4/edit?usp=sharing&ouid=109782802365303451748&rtpof=true&sd=true
Impact: There are, of course, more impacts. Although one of the main ones is the impact on the environment, the other and more interesting for companies are finances, because they allow you to save large sums with this solution.
Feasibility: We are not worthy to count individual finances, but I will divorce a small idea. For a quality solution, a set of drones would cost about 500,000 CZK and 500,000 CZK for software, example for year. One helicopter flight costs 100,000 CZK and I don't count how many times it has to fly. Add to that wages per employee.
What you built: We have built program for recognition. Code is in directory below: https://drive.google.com/drive/folders/1S_yXiDlj8wpvCQHACzLxtf4KEquKTh-Q?usp=sharing
What you had before: We have used TensorFlow library, otherelse thing is our.