Table 3 IOU and F1 score averaged over the five cross-validation fold of VegAnn computed at the dataset and image-level.

From: VegAnn, Vegetation Annotation of multi-crop RGB images acquired under diverse conditions for segmentation

Model

Encoder

IOUimage

F1image

IOUdataset

F1dataset

Unet

Resnet34

86.0 ± 1.2

91.0 ± 0.9

89.7 ± 1.4

94.5 ± 0.8

Unet

Resnet50

86.3 ± 0.8

91.4 ± 0.8

90.0 ± 0.5

94.8 ± 0.3

DeepLabV3

Resnet34

85.4 ± 0.6

90.8 ± 0.5

89.5 ± 0.2

94.5 ± 0.2

DeepLabV3

Resnet50

84.9 ± 0.8

90.5 ± 0.7

89.0 ± 0.8

94.2 ± 0.5

Back to article page