Table 6 Comparison of results of different algorithms.
From: Lightweight insulator target detection algorithm based on improved YOLOX
Model | Backbone | Params/M | Precision(%) | Recall(%) | mAP | FPS/f/s |
|---|---|---|---|---|---|---|
YOLOv3 | Darknet53 | 61.95 M | 96.43% | 72.32% | 90.83% | 72.288 |
YOLOv5 s | CSPDarknet53 | 47.05 M | 93.67% | 74.00% | 91.83% | 65.711 |
SSD | VGG16 | 23.61 M | 93.50% | 58.08% | 76.3% | 63.77 |
Faster-RCNN | VGG16 | 136.68 M | 64.10% | 99.98% | 98.60% | 47.254 |
YOLOX | CSPDarknet | 8.938 M | 91.45% | 96.00% | 96.16% | 56.08 |
YOLOv4-Tiny | Darknet19 | 5.87 M | 93.10% | 81.00% | 90.62% | 177.99 |
YOLOv7-Tiny | ELAN-Net | 6.227 M | 94.58% | 90.57% | 92.74% | 129.59 |
YOLOv8-n | CSPDarknet | 3.157 M | 91.74% | 97.00% | 96.86% | 111.26 |
YOLO11 s | C3 K2 | 9.40 M | 91.60% | 90.8% | 93.60% | 156.25 |
Ours | D-CSPDarknet | 4.164 M | 98.81% | 100.00% | 97.46% | 90.37 |
