Table 1 Summary of evaluation metrics for models trained on the HTS and N/NTS

From: VespAI: a deep learning-based system for the detection of invasive hornets

Training data

Description

Vespa velutina

Vespa crabro

mAP

F1 Score

mAP

F1 Score

HTS

Hornet training subset consisting primarily of hornets with few non-target insects

0.941

0.989 (0.668)

0.922

0.981 (0.668)

H/NTS

Hornet/non-target training subset consisting of hornets and a broad range of non-target insects

0.960

0.996 (0.194)

0.951

0.992 (0.194)

  1. Results are derived from model evaluation using the validation subset (VS) data, brackets indicate the confidence values at which F1 scores are calculated, mean average precision (mAP) values are calculated within the confidence range of 0.05–0.95.
Back to article page