Bird Flight Risk Using Google Cloud Vision Object Detection

Bird strikes on aircraft are estimated to cause 1.6 billion dollars worth of damage each year. Object detection systems can be used to mitigate bird strikes by evaluating risk before takeoff.

Comment

Inspiration

This project is inspired by our need for aircraft flight safety. In our research into the Federal Aviation Administration, we found that one of the most pressing matters involves wildlife strikes by US aircraft, and although the crashes involve a myriad of wildlife, the majority of the crashes involve birds.

What it does & how we built it

Aside from creating a Service Account and Key through Google Cloud’s website, the algorithm of the project would involve the use of the Google Application Credentials and Service Key, and having them read and link to the desired JPEG images of the bird wildlife, all saved under the same virtual environment. Next, the contents of the image would be analyzed and print the number of objects in the image, the classification of the object, and the confidence level.

Challenges we ran into

There were many challenges faced during the project, which include the setup of Google Vision and ensuring that it is installed into the Visual Studio Code text environment. Another challenge was learning about the Pillow library for Python and how to integrate that into the coordinates supplied by the objects detected in the images in order to draw the borders around the desired objects.

Accomplishments that we're proud of

We are happy about the progress that has been made in using object detection for still images. So far we can successfully classify different animals using Google Vision API.

What we learned

This project opened avenues into learning about Google Cloud Vision’s API, and allowed us to brainstorm to build our project using the Google Cloud Vision.

What's next for Bird Flight Risk Using Google Cloud Vision Object Detection

Overall, this was a very rewarding experience and motivates us to not only complete the project but to expand and improve upon the program to platforms, such as using Google Cloud’s API Live Object Detection and Tracking to detect objects in live streams, which would be ideal for pilots who need to scan their line of vision in real-time flight.

Built With

Share this project:

Updates