Mastering Computer Vision with TensorFlow 2.x

Generating annotation files

This step involves creating bounding box coordinates for each object in each image in the dataset. This bounding box coordinate is typically represented by four parameters: (x,y) to determine initial location and the width and height. The bounding box can be expressed as .xml or in .txt form. This coordinate file is also known as an annotation file. Follow these steps to complete this section:

1. A lot of image annotation software applications are being used for labeling images. We have already covered VGG Image Annotator in Chapter 3, Facial Detection Using OpenCV and CNN, during facial key-point detection. In Chapter 11, Deep Learning on Edge Devices with CPU/GPU Optimization, we will cover CVAT tool for automated ...