Skip to content

Fix new publication format #77

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dpascualhe merged 1 commit into from
May 15, 2026
Merged

Fix new publication format #77

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 40 additions & 14 deletions ...ation_of_visual_semantic_segmentation_models_for_off_road_autonomous_driving.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,37 +1,64 @@
---
permalink: /publications/2026/cross_dataset_evaluation_of_visual_semantic_segmentation_models_for_off_road_autonomous_driving
title: "Cross-dataset Evaluation of Visual Semantic Segmentation Models for Off-road Autonomous Driving"


layout: single

classes: wide
---

<p style="text-align: center; font-weight: bold;">Neurocomputing, 2024</p>
<style>
/* Centers the main page title generated by the Jekyll layout */
h1, .page__title {
text-align: center !important;
}
</style>

<p style="text-align: center"><a href="https://sergiopaniego.github.io/">David Pascual-Hernández<sup>1</sup></a>, <a href="https://sergiopaniego.github.io/">Sergio Paniego<sup>1</sup></a>, <a href="https://servicios.urjc.es/pdi/ver/roberto.calvo">Roberto Calvo-Palomino<sup>1</sup></a></p>, <a href="https://servicios.urjc.es/pdi/ver/inmaculada.mora">Inmaculada Mora-Jiménez<sup>1</sup></a></p>, <a href="https://gsyc.urjc.es/jmplaza/">Jose Maria Cañas-Plaza<sup>1</sup></a></p>
<p style="text-align: center; font-weight: bold;">Expert Systems with Applications, 2026</p>

<p style="text-align: center">
<a href="https://dpascualhe.github.io">David Pascual-Hernández<sup>1</sup></a>,
<a href="https://sergiopaniego.github.io">Sergio Paniego<sup>1</sup></a>,
<a href="https://servicios.urjc.es/pdi/ver/roberto.calvo">Roberto Calvo-Palomino<sup>1</sup></a>,
<a href="https://servicios.urjc.es/pdi/ver/inmaculada.mora">Inmaculada Mora-Jiménez<sup>1</sup></a>,
<a href="https://gsyc.urjc.es/jmplaza/">Jose Maria Cañas-Plaza<sup>1</sup></a>
</p>
<div class="container" style="overflow: hidden;">
<p style="text-align: center; width: 50%; float: left;">1: <a href="https://www.urjc.es/"><img src="https://upload.wikimedia.org/wikipedia/commons/thumb/8/84/URJC_logo.svg/1280px-URJC_logo.svg.png" width="40%" height="40%" alt="URJC"/></a></p>
<p style="text-align: center">1: <a href="https://www.urjc.es/"><img src="https://upload.wikimedia.org/wikipedia/commons/thumb/8/84/URJC_logo.svg/1280px-URJC_logo.svg.png" width="40%" height="40%" alt="URJC"/></a></p>
</div>
<p style="text-align: center">DOI: <a href="https://doi.org/10.1016/j.eswa.2026.132656">10.1016/j.eswa.2026.132656</a></p>


## Abstract

<p style="text-align: justify;">
Intelligent autonomous driving in off-road environments is an emerging field with great potential to impact areas such as agriculture, forestry, and rescue operations. Perception in these scenarios presents unique challenges due to the diversity of elements and weather conditions, along with the inherent ambiguity in class definitions. Consequently, off-road visual semantic segmentation datasets remain underdeveloped, roughly ten times smaller than their urban counterparts, hindering dependable performance assessment and potentially compromising the safety of autonomous systems. To address these challenges, we present a comprehensive cross-dataset evaluation of visual semantic segmentation models for autonomous off-road navigation. We propose a unified ontology that harmonizes class definitions across relevant datasets, enabling their combination for both training and testing. This approach ensures fair model comparisons and reliable assessment of generalization to unseen domains. We further benchmark models on the original datasets, analyze the impact of different ontology harmonization criteria and conversion strategies, and evaluate the trade-off between segmentation performance and computational cost. Results show that Transformer-based architectures achieve the most consistent segmentation performance across datasets. While often computationally demanding, some variants maintain real-time inference (≈12 ms) with top-tier accuracy. The unified ontology simplifies the segmentation task, yielding more reliable models and about 40% faster training convergence. Cross-dataset training further enhances generalization, improving mean IoU by up to +20% on RUGD and +13% on WildScenes compared to RELLIS-3D-only training. Overall, this study provides valuable insights for developing robust perception modules for off-road autonomous vehicles.
</p>

<div style="display: flex;justify-content: space-around;margin-bottom: 20px;">
<div style="width: 45%;">
<img src="https://ars.els-cdn.com/content/image/1-s2.0-S0957417426015691-gr2_lrg.jpg" frameborder="0" allowfullscreen></img>

<div style="display: flex; justify-content: space-between; align-items: flex-start; margin-bottom: 20px; gap: 20px;">
<div style="width: 40%; display: flex; flex-direction: column; gap: 25px;">
<div style="width: 100%;">
<img src="https://ars.els-cdn.com/content/image/1-s2.0-S0957417426015691-gr4_lrg.jpg" alt="Examples of the ontology conversion proposed for enabling cross-dataset evaluation." style="width: 100%; height: auto;">
<p style="font-size: 0.85em; color: #555; margin-top: 8px; text-align: justify; line-height: 1.4;">
Examples of the ontology conversion proposed for enabling cross-dataset evaluation.
</p>
</div>
<div style="width: 100%;">
<img src="https://ars.els-cdn.com/content/image/1-s2.0-S0957417426015691-gr2_lrg.jpg" alt="Overview of our cross-dataset training and evaluation pipeline." style="width: 100%; height: auto;">
<p style="font-size: 0.85em; color: #555; margin-top: 8px; text-align: justify; line-height: 1.4;">
Overview of our cross-dataset training and evaluation pipeline.
</p>
</div>
</div>
<div style="width: 45%;">
<img src="https://ars.els-cdn.com/content/image/1-s2.0-S0957417426015691-gr11_lrg.jpg" frameborder="0" allowfullscreen></img>

<div style="width: 50%;">
<img src="https://ars.els-cdn.com/content/image/1-s2.0-S0957417426015691-gr11_lrg.jpg" alt="mIoU vs. average inference time per image." style="width: 100%; height: auto;">
<p style="font-size: 0.85em; color: #555; margin-top: 8px; text-align: justify; line-height: 1.4;">
mIoU vs. average inference time per image. Models trained on the combined RELLIS-3D and GOOSE train datasets, and evaluated on RUGD (a) and WildScenes (b) complete datasets. Bubble size represents the number of parameters for each model. Labels indicate model names. Bold labels highlight Pareto-optimal models.
</p>
</div>
</div>

## Materials

## Materials
<div class="container" style="overflow: hidden;">
<div style="width: 33%; float: left;margin-bottom: 20px; text-align: center;">
<a href="https://doi.org/10.1016/j.eswa.2026.132656">
Expand All @@ -55,7 +82,6 @@ Intelligent autonomous driving in off-road environments is an emerging field wit


## Citation

```
@article{pascual2026cross,
title={Cross-Dataset Evaluation of Visual Semantic Segmentation Models for Off-Road Autonomous Driving},
Expand Down
2 changes: 1 addition & 1 deletion _pages/publications/publications.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ Journals, congress papers and research publications can be found below:

# 2025

* [Deep Learning-Based Semantic Segmentation of LiDAR Point Clouds for Autonomous Driving in Unstructured Environments, Proceedings of the XXV International Workshop on Physical Agents (WAF). Félix Martínez, David Pascual-Hernández, Daniel Borja Fernández, Inmaculada Mora Jiménez, José María Cañas]
* Deep Learning-Based Semantic Segmentation of LiDAR Point Clouds for Autonomous Driving in Unstructured Environments, Proceedings of the XXV International Workshop on Physical Agents (WAF). Félix Martínez, David Pascual-Hernández, Daniel Borja Fernández, Inmaculada Mora Jiménez, José María Cañas.

# 2024

Expand Down