Put Your MASKON

“Put your mask on” Introduction Since March of last year, we have all been affected by a pandemic that marked our lives forever, Covid-19. This pandemic changed the way we interact with other people, the way we travel, the way we study, and even the way we love. It has certainly been an exceedingly difficult process of adaptation, but we are aware that with the help of everyone it is possible to return to the reality we so desire. That's why we've created MASKON, a mask detecting product created in the Tinker Lab with recycled materials and a Rasberry Pi to help stop this terrible pandemic that's ended many people's lives. This product helps to recognize if people are wearing their mask properly. According to the FDA: “Masks may help prevent people who have COVID-19 from spreading the virus to others. The CDC recommends people wear face masks in public settings, especially when other social distancing measures are difficult to maintain. Wearing a face mask may limit exposure to respiratory droplets and large particles and may help prevent people who have COVID-19 from spreading the virus”. If we encourage everyone to use their mask properly, we will create a safer environment for all. Methodology Base The base was created using 8020 tubing, since they provided good support while also being lightweight aluminum. We didn’t have very long pieces of 8020, so we had to create some aluminum brackets to join the small sections; we made the most out of what we had. To support the weight of the components, we added a steel bar on the back to act as a counterweight. The 8020 also provides an easy way to quickly adjust the device for different ranges. Components Panel We used a plywood sheet to secure the components to the base. The Raspberry Pi, Camera, and Touchscreen are supported by a custom-made acrylic bracket, which alleviates the weight off of the Touchscreen. We also added a wooden box to contain extra masks if someone needs them. Technical Face-mask Detection The code is written entirely in Python and the machine learning code was adapted from an online source. First, it takes an image of the user. To detect faces in the image, the program uses the Haar Cascade Classifier from the OpenCV library. It then analyzes the faces using TensorFlow and a trained model. The model is trained using a vast and diverse dataset we found of random people with and without masks. The result is passed back to the script to notify the user. Note: The images are not saved on the device, as they are overwritten every time it takes another image. The device also has the capability of running completely offline, making it a safe and secure program. Notifying the User To notify the user, we use an RGB LED Strip with specific colors for each result code: 0 = No Mask - Red 1 = Masked - Green 2 = No faces detected - White The LEDs are controlled by an Arduino MKR Wi-Fi 1010 which receives commands from the Raspberry Pi through the serial port. Note: We intend to add other notification methods in the future, with the purpose of making the device as inclusive as possible and more effective. These methods may include audio responses, verbal screen responses, among others. Results & Future Work Maskon is a product that could easily be built at home if you have the right equipment; all you need is a computer with a camera, access to the internet, creativity, and to enjoy the process. Nevertheless, having access to a manufacturing environment facilitates the process and helps create a more sophisticated product. If you want to create a movable and solid platform, then this environment is necessary. You should start by creating rigid base that could support outdoor winds so that your Maskon could be used on any occasion and be practical for society. For example, we joined different 8020 tubes with screws and used a steel bar to provide the necessary support. Then you may consider building a holder where you can accommodate your display and the camera in a reasonable height, so that this machine works for almost any person no matter the height or you may also consider building an adjustable holder that could easily be moved. For the technical parts of this assembly, you don’t need a manufacturing environment since most of this work could be made from the commodity of your house. In a near future, we would like to make our product more easily manufactured at a larger scale by reducing its size, its weight, making custom brackets that could be machined, and although the device as of now is not awfully expensive we would consider making it more cost efficient so that any person could have the privilege to build its own Maskon. Problems encountered Some of the problems we faced during the development of Maskon were primarily because of lack of materials, but we made the most out of what we had. For instance, we spent a long-time making aluminum brackets to join small 8020 pieces and make sure it was rigid. We could have avoided these issues by having a longer piece of 8020 cut to length. Likewise, we spent a long time looking for screws that would work with the 8020. With so many brackets, we quickly ran out of T nuts and had to find alternatives. Regarding the touchscreen support bracket, we really had to engineer it, so it was sturdy while not putting stress on the touchscreen. We used an acrylic piece, along with a piece of wood to support it to the panel. The holes on the bracket didn’t align properly, so we had to do some adjusting so it fitted well with the screen. However, if we had the time, we would have liked to design a 3D printed or machined bracket that would have done this in a better and easier way with precise measurements. While mounting the components panel on the base, we initially didn’t get it level because of the holes we made. We fixed this by giving the holes some horizontal play with the mill (this can be seen on the video). Moreover, getting the Raspberry Pi to communicate with the LED Strip was a hassle. We initially intended to use an Adafruit Flora, but it didn’t communicate well through serial with the Raspberry Pi. We then changed to the Arduino MKR Wi-Fi 1010, which did work. In the future, we would like to either connect the LED Strip directly to the Raspberry Pi, or use an easier method of controlling them. BOM * Plywood Sheet * Rasberry Pi 3B+ * Rasberry Pi Camera V2 * Arduino MKR Wi-Fi 1010 * Rasberry Pi Touchscreen * 8020 tubing * LED Strip * Screws (M3 & M5) * Masks * Wood Glue * Weight (Steel bar) * Extension cords * Keyboard * Acrylic * AWG 20 cables * 5V power supply CAD assembly + parts drawing