[#AiDIGEST] Three Best Practices - Image Data Annotation

1. Label the whole object. It is better to include a buffer than to cut out part of the object you are labeling.
2. If another object obscures the object, label it as if you can see the full thing. This helps the model to understand the actual size and area of the object.
3. If an object is partially out of the frame, label it anyway.

One of the cool tools available out there is LabelImg which will make a Data Scientist's work a breeze, especially during the exploratory data analysis stage.