Table 1 Comparison of model performance in terms of task type, accuracy, and real-time inference speed (FPS). Highlights YOLOv9 and DeepLabv3+ as the top-performing models.

From: Cascade drive: a unified deep learning framework for multi-featured detection and control in autonomous electric vehicles on unstructured roadways

Models

Feature

Accuracy

FPS

VGG16

Image classification

78.87%

1.6

YOLOv9

Object detection

92.3%

6

YOLOv9

Pothole detection

83%

5.5

SSD

Object detection

92%

5.5

DeepLabv3+

Lane detection

97.26%

6

UNet

Lane detection

91.2%

2.5