> ## Content Index
> Fetch the complete content index at: https://jeroenoverschie.nl/llms.txt
> Use this file to discover other available public pages before exploring further.

# A fusion between a deep-learning and a filter-based approach for road marker detection
- URL: https://jeroenoverschie.nl/bsc-thesis-road-marker-detection/
- Published: 2019-08-23T12:00:00.000Z
- Updated: 2026-10-30T09:14:27.000Z
- Description: Fusing the outputs of a supervised deep-learning and an unsupervised filter-based road marker detector into a new supervised model, reaching 95% accuracy.
- Author: Jeroen Overschie
- Tags: Research, AI & ML, #code, #pdf

By Jeroen G. S. Overschie. Bachelor's thesis, **University of Groningen**, supervised by dr. G. Azzopardi, in assignment for Kaios.ai. Published 23 August 2019.

[Code](https://github.com/dunnkers/bsc-thesis) · [PDF](https://github.com/dunnkers/bsc-thesis/blob/master/report.pdf)

## Abstract

In this study, a fusion between two existing algorithms for detecting road markings is proposed. Considered are two existing algorithms, one supervised, deep-learning algorithm, and another unsupervised filter-based algorithm. The proposed fusion algorithm takes the output images of both approaches in order to train a new model using supervised learning. Under different conditions of downscale factor, classifier and feature-vector composition, the fusion algorithm managed to obtain an accuracy level of 95%.

[GitHub - dunnkers/bsc-thesis: BSc Thesis - A fusion between a deep-learning and a filter-based approach for road marker detectionBSc Thesis - A fusion between a deep-learning and a filter-based approach for road marker detection - dunnkers/bsc-thesis![](https://jeroenoverschie.nl/content/images/icon/favicon-ea24923b-b521-46f6-8b01-35eb05fc3bb7.png)GitHub![](https://jeroenoverschie.nl/content/images/2026/10/bsc-thesis-og.png)](https://github.com/dunnkers/bsc-thesis)