Problem with existing approach

Although the MobileNet SSD is fast and gives us good results, it still can't identify classes such as cup, pen, and so on. So, we need to use the pre-trained model that has been trained on a variety of objects. In this upcoming iteration, we need to use the pre-trained model, for example, the TensorFlow object detection API, which will able to identify the different objects compared to the baseline approach. So now, let's look at how we will optimize the existing approach.