Object Detection with Machine Learning on a RaspberryPi
This project uses a Raspberry Pi camera to capture an image and detect the object that is in view. You can view this process on an HTML website. In addition the page gives a QR code in which you an scan with your phone and learn more about it.
| Engineer | School | Area of Interest | Grade |
|---|---|---|---|
|
Darshan M. |
Amador Valley High School |
Computer Science |
Rising Senior |
FIRST MILESTONE
During the third week, I used HTML and CSS to build a website that displays the live video feed of what the camera sees. The website provides the same instructions as the Visual Studio Code interface, meaning that it tells the user to cover the camera in order to take a picture. Once the image is taken, the final image that was captured will show up. I included a button labeled as “Click me to reveal object” that you can press. Once pressed it changes the title to “The possible object is: “ and the text beneath to whatever object it detected, In addition, the image that displayed the video feed changes to a QR code about the object and when scanned with an iphone leads the user to a wikipedia link.
During the first week, I tried to connect the raspberry pi wirelessly through puTTy since I didn’t have the suitable cable yet. When I received the cable I was able to use the raspberry pi terminal through Visual Studio Code or VNC viewer. Using VNC viewer I was able to power up the camera that I connected to my raspberry pi. And I used Visual Studio Code to run python scripts. I used Visual Studio Code to run machine learning models and the camera.
SECOND MILESTONE
During the second week, I used Visual Studios Code to run python to capture an image through the raspberry pi. I then ran the image through a machine learning model to identify what object it was. Then I modified it so that the python code captures the image contsantly thus acting as a live feed. Then I tweaked the script such that the camera would take a picture if the camera was covered by detecting darkness. This image would be run through a machine learning code and would spit out the most likely object. I then modified it further such that it gives a QR Code that you can scan and direct the user to a Wikipedia link about the object.
FINAL MILESTONE
During the third week, I used HTML and CSS to build a website that displays the live video feed of what the camera sees. The website provides the same instructions as the Visual Studio Code interface, meaning that it tells the user to cover the camera in order to take a picture. Once the image is taken, the final image that was captured will show up. I included a button labeled as “Click me to reveal object” that you can press. Once pressed it changes the title to “The possible object is: “ and the text beneath to whatever object it detected, In addition, the image that displayed the video feed changes to a QR code about the object and when scanned with an iphone leads the user to a wikipedia link.