I was so fortunate to have a summer internship as a data scientist at Q-Free, Netherlands in 2019. There I was researching the use of neural networks for Automatic License Plate Recognition. The task was to recognize the color of License Plates. This is quite a challenging task as the light conditions and picture quality is of such importance. In the yellow light from a lamppost, a white license plate can seem yellow. For Automatic License Plate Recognition, it is an emphasis on quick inference, due to the high speed of cars. Therefore it was decided to use the neural network architecture MobileNetV2, which is a lightweight architecture made for quick predictions. One of the key lessons from the experience was the importance of structure and good scientific methods.