- Implementation of depth map fusion significantly improved segmentation accuracy.

- Achieved higher pixel-accuracy, Intersection over Union (IoU), and Dice Coefficient compared to the base model.

- The U-Net architecture proved highly efficient for road segmentation tasks requiring precise localization.

- Its design facilitated detailed and accurate segmentation outputs by effectively using skip connections and a deep learning framework.

- Employed data augmentation techniques like horizontal flipping to enhance model training.

- Augmentation helped the model generalize better across different driving contexts, reducing overfitting.

- Fused model outperformed the standard model in all key metrics, underscoring the benefit of incorporating depth data.

- Encountered difficulties in classifying lanes with distinct markings due to high contrast.

- Proposed solutions include expanding the dataset to encompass a wider variety of lane markings and traffic scenarios.

- Employed 3D plotting to visually demonstrate the model's ability to detect and plot road surfaces.

- The 3D point cloud visualizations highlighted the practical application of the model in real-world scenarios.