IGLOSS: Image Generation for Lidar Open-vocabulary Semantic Segmentation

Nermin Samet, Gilles Puy, Renaud Marlet
4/1/2026
cs.CV

Official code: valeoai/IGLOSS (5 stars, last commit 2026-04-28).

Abstract

This paper presents a new method for the zero-shot open-vocabulary semantic segmentation (OVSS) of 3D automotive lidar data. To circumvent the recognized image-text modality gap that is intrinsic to approaches based on Vision Language Models (VLMs) such as CLIP, our method relies instead on image generation from text, to create prototype images. Given a 3D network distilled from a 2D Vision Foundation Model (VFM), we then label a point cloud by matching 3D point features with 2D image features of these prototypes. Our method is state-of-the-art for OVSS on nuScenes and SemanticKITTI. Code, pre-trained models, and generated images are available at https://github.com/valeoai/IGLOSS.

View on arXivView PDF

Code Implementations(1)

valeoai/IGLOSSOfficial25%

Official PyTorch implementation of IGLOSS Image Generation for Lidar Open-vocabulary Semantic Segmentation

50Shell, PythonApr 1, 20265 months ago
2d-to-3ddistillationimage-generationlidar-point-cloudopen-vocabulary-segmentation+1 more

Cite this paper

@article{samet2026igloss,
  title  = {IGLOSS: Image Generation for Lidar Open-vocabulary Semantic Segmentation},
  author = {Nermin Samet and Gilles Puy and Renaud Marlet},
  year   = {2026},
  eprint = {2604.01361},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2604.01361v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2604.01361.svg)](https://codeofpaper.com/paper/2604.01361)

Discussion