Fig 1.
Overall schematic of deep learning based polyp detection.
Table 1.
The details of the selected datasets in polyp detection.
Fig 2.
The brief structure of implemented YOLOv5 object detection model.
Fig 3.
(a) The AP at IoU 0.5 in training process; and (b) the AP at IoU 0.5:0.95 in training process with Adam optimizer.
Fig 4.
(a) The AP at IoU 0.5 in training process; and (b) the AP at IoU 0.5:0.95 in training process with SGD optimizer.
Fig 5.
(a) The ground truth of the selected images; and (b) the predicted results of the selected images in GIANA2017-T.
Table 2.
Results of model trained with Adam optimizer on test set GIANA2017-T.
Table 3.
Results of model trained with SGD optimizer on test set GIANA2017-T.
Table 4.
Comparison of models on test set GIANA2017-T.
Table 5.
The list of some representative predictions in the test set.
Table 6.
Comparison of models on test set CP-CHILD-AT.
Fig 6.
(a) The ground truth of the images; and (b) the predicted results of the images in CP-CHILD-AT.
Fig 7.
The quality management and control framework in model development.