Live classification must have shown some limitation of the Mobilenet model, and the method the ML classification works. We will improve on that by adding object detection using ml5 object detector method that uses the COCO and YOLO models
The learning was on the usage of P5's function to bring the bounding rectangles around the objects. The model predicts the objects, along with their location and the approximate bounding rectangles around them. It is provided as JSON file which can then be referred for creating the rectangles, the labels etc