Mastering Computer Vision with TensorFlow 2.x

Processing key-point data

The VGG annotator tool generates an output CSV file that needs to be preprocessed twice to generate the (x,y) coordinates for each of the 16 key points for each image. This is a very important concept for large data processing that you can use for other computer vision tasks for three main reasons:

• Our Python code will not search a directory for lots of image files directly but will search the input CSV for the data path.
• For each CSV file, there are 16 corresponding key-points that need to be processed.
• This is alternative to using a Keras ImageDataGenerator and flow_from_directory method to look through each file in a directory.

To clarify this content, this section is broken down into the following two subsections: ...