Conceptual Cognitive Modeling for Fine-Grained Annotation Quality Assessment of Object Detection Datasets

Table 2

Annotation accuracy of human annotations for the downsampled UTS dataset.

Class	Flat accuracy of label		Hierarchical accuracy of label
	Precision	Recall	Precision	Recall
Hatchback	0.79 ± 0.02	0.55 ± 0.14	0.92 ± 0.01	0.81 ± 0.07
Sedan	0.58 ± 0.08	0.78 ± 0.12	0.86 ± 0.03	0.88 ± 0.07
Minibus	0.93 ± 0.10	0.57 ± 0.33	0.95 ± 0.06	0.70 ± 0.22
SUV	0.20 ± 0.06	0.27 ± 0.10	0.69 ± 0.03	0.74 ± 0.04
MPV	0.19 ± 0.14	0.26 ± 0.14	0.62 ± 0.14	0.72 ± 0.08
Pickup	0.57 ± 0.41	1 ± 0	0.72 ± 0.27	1 ± 0
On average	0.55 ± 0.11	0.57 ± 0.07	0.79 ± 0.07	0.81 ± 0.05