Showing 1–1 of 1 results for author: Breiter, M

Explaining YOLO: Leveraging Grad-CAM to Explain Object Detections

Authors: Armin Kirchknopf, Djordje Slijepcevic, Ilkay Wunderlich, Michael Breiter, Johannes Traxler, Matthias Zeppelzauer

Abstract: We investigate the problem of explainability for visual object detectors. Specifically, we demonstrate on the example of the YOLO object detector how to integrate Grad-CAM into the model architecture and analyze the results. We show how to compute attribution-based explanations for individual detections and find that the normalization of the results has a great impact on their interpretation. We investigate the problem of explainability for visual object detectors. Specifically, we demonstrate on the example of the YOLO object detector how to integrate Grad-CAM into the model architecture and analyze the results. We show how to compute attribution-based explanations for individual detections and find that the normalization of the results has a great impact on their interpretation. △ Less

Submitted 22 November, 2022; originally announced November 2022.

Journal ref: Proceedings of the Workshop of the Austrian Association for Pattern Recognition 2021