Figure2
From: A deep learning-based system for accurate detection of anatomical landmarks in colon environment
Figure 2. Workflow of the proposed landmark detection system. The pre-processing module rejects interference frames in input data. The pre-processed data are divided into training, validation, and test sets. The detection model based on ResNet-101 is trained and validated using the training and validation sets. After passing the test data into the trained model, the model outputs intermediate detection results indicating whether each frame should be classified as positive or negative. Finally, the post-processing module identifies the incorrectly predicted frames and reassigns them back to the correct class. P-N gap: Positive-and-negative gap.
