Table 1 Model prediction performance comparison of object detection with deep learning.

From: Heat-vision based drone surveillance augmented by deep learning for critical industrial monitoring

Model

mAP50

mAP50-95

Params

FLOPS

YOLOv5n35

0.973

0.742

1.9 M

4.5 G

YOLOv5s35

0.970

0.767

7.5 M

16.5 G

YOLOv5m35

0.971

0.756

21.2 M

49 G

YOLOv8n36

0.977

0.779

3.2 M

8.7 G

YOLOv8s36

0.981

0.799

11.2 M

28.6 G

YOLOv8m36

0.981

0.804

25.9 M

78.9 G

NanoDet-Plus37

0.959

0.716

1.17 M

1.52 G

Mask R-CNN-1017

0.919

0.716

55 M

937 G

Mask R-CNN-507

0.916

0.711

36 M

890 G

Back to article page