Testing and Results

Brand and Product Detection

The results of the testing were overall successful with only one product that was incorrectly identified. The Adobo blue bottle should have given an output of “this is abodo without pepper” however it gave the output “this is old bay blackened seasoning”. We believe that this is because it did not detect the adobo logo correctly, possibly due to lighting, and when running the OCR, it read the word seasoning and selected blackened seasoning from our product list. Overall, with a success rate of 93.33% we believe that the results of the real-world testing were a success.

Error Message

There are 2 different error messages that will output depending on the error that occurs. The first error message is for when there is no hand detected in the image. When this event occurs, the project will not be able to determine a selection. Therefore, the error message containing the string of text “No hand detected” is sent through the bone conducting headphones to inform the user. The project was tested 5 times with the user’s hand kept out of view.

No Hand Error Message

Nothing Detected Error Message

Battery Life

For testing the battery life of the project, we first charged the power supply fully as indicated by a green light that illuminates when the device is fully charged. Then we turned the camera on and let it run at maximum resolution for the duration of each test. A camera was placed in front of the raspberry pi and a clock to monitor the project during each trail run.

Operating Temperature

During the 5 trial runs ran the project and checked the CPU temperature using the shell command “vcgencmd measure_temp” we then recorded the maximum temperature that the CPU reached during each trial run.

Comfort

To test the comfort of the hat we allowed 5 people to try on the hat and give us their honest feedback. We asked each person to try them on and give a score to each category on a scale of 1 to 5, 5 being the best score and 1 being the worst.