Table 4 Base YOLOv8 models performance and parameters on test normal-case dataset and test hard-case dataset.
From: Enhancing the YOLOv8 model for realtime object detection to ensure online platform safety
Train Dataset | Test Dataset | Models | Precision | Recall | mAP50 | mAP50-90 | Parameters |
|---|---|---|---|---|---|---|---|
train normal Dataset | test normal Dataset | YOLOv8n | 0.72 | 0.68 | 0.71 | 0.55 | \(3.007 \textrm{M}\) |
YOLOv8s | 0.73 | 0.65 | 0.74 | 0.58 | \(11.12 \textrm{M}\) | ||
YOLOv8m | 0.72 | 0.68 | 0.77 | 0.62 | \(25.84 \textrm{M}\) | ||
test hard Dataset | YOLOv8n | 0.65 | 0.58 | 0.55 | 0.43 | \(3.007 \textrm{M}\) | |
YOLOv8s | 0.67 | 0.57 | 0.65 | 0.46 | \(11.12 \textrm{M}\) | ||
YOLOv8m | 0.66 | 0.61 | 0.69 | 0.55 | \(25.84 \textrm{M}\) | ||
train ( normal Dataset \(\cup\) hard Dataset) | test normal Dataset | YOLOv8n | 0.75 | 0.66 | 0.72 | 0.58 | \(3.007 \textrm{M}\) |
YOLOv8s | 0.70 | 0.68 | 0.79 | 0.63 | \(11.12 \textrm{M}\) | ||
YOLOv8m | 0.79 | 0.75 | 0.85 | 0.68 | \(25.84 \textrm{M}\) | ||
test hard Dataset | YOLOv8n | 0.68 | 0.67 | 0.68 | 0.55 | \(3.007 \textrm{M}\) | |
YOLOv8s | 0.72 | 0.68 | 0.75 | 0.63 | \(11.12 \textrm{M}\) | ||
YOLOv8m | 0.73 | 0.70 | 0.76 | 0.57 | \(25.84 \textrm{M}\) |
