Search and Rescued

Inspiration

After a natural disaster, many victims become stranded. While aerial searches are helpful, it can be sometimes hard to spot victims from such a great altitude. Having a second pair of "eyes" would be very useful as it could save a victim's life. That is where our hack comes in :)

What it does

Search and Rescued is an application that is mounted onto an aerial vehicle that will give a live video stream. Using object detection, Search and Rescued will be able to spot a stranded victim that a human might have missed for a variety of reasons. After receiving this feedback through object detection, a rescue team can save the life of a victim after a natural disaster or loss.

How I built it

With blood, sweat, and tears.

We used TensorFlow, an object detection API with a COCO dataset, and coded everything in python.

Challenges I ran into

There were many, many, many challenges.

Accomplishments that I'm proud of

Getting the object detection to work with a live video stream

What I learned

Using the command prompt, opencv, python, tensorflow, object detection, and much more!

What's next for Search and Rescued

Given the power of the application and the hardware limitations that we had, the next step would be to get the necessary hardware that would allow an increase in the number of frames per second. After, we could add a GPS dongle that would indicate the exact location of a detected person.

Built With

• dragon-board
• opencv
• python
• raspberry-pi
• tensorflow