Evaluating object detector ensembles for improving the robustness of artifact detection in endoscopic video streams

TL;DR

This paper presents an ensemble deep-learning approach combining YOLOv4 and Yolact to improve artifact detection in endoscopic videos, achieving higher accuracy without sacrificing real-time performance.

Contribution

It introduces a novel ensemble strategy for combining two one-stage detectors, enhancing robustness and accuracy in endoscopic artifact detection.

Findings

Ensemble approach outperforms individual models in mean average precision

The method maintains real-time detection capabilities

Surpasses previous state-of-the-art results on the dataset

Abstract

In this contribution we use an ensemble deep-learning method for combining the prediction of two individual one-stage detectors (i.e., YOLOv4 and Yolact) with the aim to detect artefacts in endoscopic images. This ensemble strategy enabled us to improve the robustness of the individual models without harming their real-time computation capabilities. We demonstrated the effectiveness of our approach by training and testing the two individual models and various ensemble configurations on the "Endoscopic Artifact Detection Challenge" dataset. Extensive experiments show the superiority, in terms of mean average precision, of the ensemble approach over the individual models and previous works in the state of the art.