Summary

In this chapter, you got a deep overview of various object detector methods and practical methods of training an object detector using your own custom image from start to finish. Some of the key concepts learned are how to work with Google Cloud to evaluate an object detector, how to use labelImg to create an annotation file, how to link Google Drive to a Google Colab notebook to read files, how to generate a TensorFlow tfRecord file from .xml and .jpg files, how to start a training process and monitor readings during training, how to create TensorBoard to observe training accuracy, how to save a model after training, and how to perform inference with the saved model. Using this methodology, you can select your object class and create an object detection model for inference. You also learned various techniques for object tracking, such as Kalman filtering and neural network...