8. Sources of Error

One concern of the neural network would be an increase in percentage error while applying the model to real-life situations. As all of the images used during training of the neural network were mined from the web, a concern of mine would be the network failing to correctly identify the vector carrier in everyday situations, i.e. in imperfect lighting, unclear focus, moving vector carrier, multiple objects in the image, etc. A way to combat this would be to deploy the model prematurely and to use the identifying pictures sent in by the users in imperfect situations in the model training itself. A solution for multiple objects in the picture is simple; use the pretrained weights and dataset on ImageNet and, using bounding boxes, identify what the other objects in the picture are.