When you are annotating, you have to create a bounding box around the car to tell the model, "Okay, this is where the car is." You can use open source and third parties to work on this platform and they allow you to zoom in on the image, so you can properly mark the coordinate. However, Azure Custom Vision does not allow you to zoom in on the image. Also, once you upload an image, you do not have the provision to download it or to search the file by file name, all that facility is not available.
