The problem componentAR solves

1. We present an approach that makes learning about electronic gadgets and their little components more engaging and entertaining at a time when education is all about theoretical and bookish information.
2. It assists folks who are interested in learning more about the components that make up a device.
3. Our project is an excellent example of using technology to make education more interesting. This model is not only for students to utilize, but it may be used by anyone interested in knowing more about electronics.

Description-

When the app first launches, you'll be given the choice to scan the device you wish to learn more about. Whether it's a smartphone or a laptop, our machine learning algorithm will detect it. Our machine learning model can currently recognise these two devices because we only have 3D models for 2-3 laptops and mobile phones owing to a lack of time. Following the device's detection, a 3D model of the device will be augmented, allowing the user to see all of the components employed inside it. For object detection, which is the detection of an electronic gadget, we use the yolov5 and yoloX models. We enhance the relevant electrical device in its 3d model once the gadget has been identified so that the user can explore it. We're utilizing web scraping to get information about the device, such as color and features. Yolo models have been converted to Onnx object detection models so that they can be utilized in continuous feed and operated in a C sharp script. This information is retrieved through a QR code on the device that displays information on the electronic components; the benefit of this is that it can display accurate specs even after the device has been modified and updated by a qualified person. Not only that, but the user can also see the provided details about these device components and learn more about them by selecting the option to see more details.

Challenges we ran into

1. To keep the application's size down, we had to make all of the 3D models ourselves.
2. It was also difficult to combine the ML model and unity.
3. Adjusting the 3d model to fit the user's device frame size so they could explore the entire device was a difficult task.
4. We used yolox and yolov5 models but Unity doesn't support them, therefore we had to convert them to an onnx object detection model.
5. We had to prepare the datasets ourselves to keep the size down.