Fig. 7From: Semantic segmentation and photogrammetry of crowdsourced images to monitor historic facadesThe loss plot for three one-stage models and two-stage model in the training phaseBack to article page