The main reason for coming up with self-driving cars is to enhance the safety of cars on the road by improving and reducing the occurrence of traffic accidents. Self-driving cars rely on advanced systems known as Advanced Driver Assistance Systems (ADASs) that help in perceiving the driving environment despite being on the road. Automobile lane detection is repeatedly considered effective in building up the recognition of the nearest lanes. Is it capable of discerning lane markings and objects on the side of the road or, topologically, it does not have to make a rolling contact? To answer this query, hence we apply road segmentation techniques that involve delineation of objects of interest along the road. In this paper, we compared two models named VGG-19 and UNET by observing their accuracy in detecting road lanes, objects, and cars. We find that the VGG-19 model achieves superior accuracy, about 40% accuracy with 93% precision in detecting complex road structures compared to the UNET model in our 95 images. This research contributes to finding a novel technique for semantic segmentation in the context of autonomous driving.
Is the VGG-19 Road Segmentation Method better than the Customized UNET Method?
2024-07-27
643289 byte
Conference paper
Electronic Resource
English
Segmentation Detection Method for Complex Road Cracks Collected by UAV Based on HC-Unet++
DOAJ | 2023
|Moth image segmentation based on improved Unet
British Library Conference Proceedings | 2022
|RAO‐UNet: a residual attention and octave UNet for road crack detection via balance loss
DOAJ | 2022
|RAO‐UNet: a residual attention and octave UNet for road crack detection via balance loss
Wiley | 2022
|