Endoscopic Artefact Detection with Ensemble of Deep Neural Networks and False Positive Elimination

Video frames obtained through endoscopic examination can be corrupted by many artefacts. These artefacts adversely affect the diagnosis process and make the examination of the underlying tissue difficult for the professionals. In addition, detection of these artefacts is essential for further automated analysis of the images and high-quality frame restoration. In this study, we propose an endoscopic artefact detection framework based on an ensemble of deep neural networks, classagnostic non-maximum suppression, and false-positive elimination. We have used different ensemble techniques and combined both one-stage and two-stage networks to have a heterogeneous solution exploiting the distinctive properties of different approaches. Faster R-CNN, Cascade R-CNN, which are two-stage detector, and RetinaNet, which is single-stage detector, have been used as base models. The best results have been obtained using the consensus of their predictions, which were passed through class-agnostic non-maximum suppression, and false-positive elimination. Index Terms Endoscopic artefact detection, Faster RCNN, Feature pyramid networks, RetinaNet