johnlockejrr/medieval-manuscript-yolov11-seg

This a collection of YOLOv11 segmentation models trained on the CATMuS Medieval Segmentation dataset which is a collection of various datasets that have SegmOnto XML markup.

YOLOv11 Model Evaluation Results

Overall Metrics

Model	mAP50-95	mAP50	mAP75
YOLOv11-Nano	0.219	0.337	0.238
YOLOv11-Small	0.242	0.385	0.242
YOLOv11-Medium	0.259	0.411	0.261
YOLOv11-Large	0.257	0.409	0.269
YOLOv11-XLarge	0.243	0.383	0.236

Per-Class mAP50-95 Metrics

Class	YOLOv11-Nano	YOLOv11-Small	YOLOv11-Medium	YOLOv11-Large	YOLOv11-XLarge
MainZone	0.831	0.836	0.846	0.854	0.865
DefaultLine	0.743	0.770	0.786	0.785	0.776
RunningTitleZone	0.024	0.074	0.058	0.076	0.097
MarginTextZone	0.129	0.169	0.199	0.177	0.164
NumberingZone	0.083	0.106	0.119	0.122	0.127
HeadingLine	0.269	0.282	0.292	0.316	0.292
DropCapitalZone	0.379	0.394	0.391	0.448	0.399
QuireMarksZone	0.051	0.139	0.162	0.184	0.184
TitlePageZone	0.000	0.000	0.000	0.000	0.000
InterlinearLine	0.018	0.041	0.146	0.085	0.114
MusicLine	0.219	0.242	0.259	0.257	0.243
StampZone	0.185	0.223	0.304	0.261	0.244
DropCapitalLine	0.198	0.214	0.191	0.186	0.039
GraphicZone	0.371	0.382	0.387	0.364	0.338
DamageZone	0.000	0.000	0.000	0.000	0.000
DigitizationArtefactZone	0.219	0.242	0.259	0.257	0.243
SealZone	0.000	0.000	0.000	0.000	0.000
MusicZone	0.219	0.242	0.259	0.257	0.243

Plot (YOLOv11-XLarge)

Project link

catmus-yolo