Week 8

02/26/2024 - 03/04/2024

Ryan: After further research it was decided to use OpenCV as the image recognition software for the project, instead of the TensorFlow library. OpenCV is an open source library that allows for computer vision, and machine learning capabilities for a device such as a Raspberry Pi. In addition to image manipulation, the program is also able to provide facial recognition capabilities.

In order to make the algorithm work, a cascade file is going to need to be used. The Haar cascade process is the chosen process for this project. The Haar cascade file works by first splitting the input image into even regions, and analyzing adjacent regions of the input image and comparing pixel intensities of the regions. Using these pixel intensities, the code will compare the input image from the camera to the images provided of the dog and look for a match. An example of the Haar comparison is shown below.

Nicolo:

The power supply module is composed of transformer, bridge rectifier, and a DC-DC converter.

Given that the microcontroller can be powered by a USB port and given that the door motor requires 3.7V DC, it was determined that a generic dual-USB power supply would be adequate for our project.

The first USB port will be used to power the Raspberry Pi microcontroller, while the second USB port will be used as a 5V DC source.

As shown in the pinout diagram below, we can obtain 5V DC by connecting the positive terminal to the RED wire, and the negative terminal to the BLACK wire.

Next, the 5V DC will be provided to a voltage regulator in order to step down the voltage to about 3.7V. For our purpose, the LM2596 was chosen[2].

A premade voltage regulating circuit including the LM2596 is available for $2.35. When a 5V DC is provided to the input of the circuit, the output voltage can be adjusted by operating the regulation knob at the top of the device. The output of this circuit will provide the 3.7V DC necessary for the door’s motor to operate.