CASWiT is a context-aware Transformer for ultra-high resolution aerial image segmentation. CASWiT achieves SOTA performances on FLAIR-HUB and URUR.
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
HEIG-VD Geomatic
We share datasets, models, and demos developed at HEIG-VD in the field of geomatics and AI for spatial data, with a strong focus on high-resolution imagery, LiDAR, photogrammetry, and multimodal learning.
Our goal is to promote reproducible research, open benchmarks, and practical baselines for real-world geospatial applications.
Research Topics
- Semantic segmentation (2D / 3D)
- Multimodal fusion (imageβLiDAR)
- Photogrammetry and georeferencing
- Large-scale and ultra-high-resolution imagery
- Geospatial AI for infrastructure mapping
- Benchmarking and reproducible baselines
Contact
For collaborations, research questions, or dataset usage:
Adrien Gressin
Professor of Photogrammetry & Data Science β HEIG-VDAntoine Carreaud
PhD candidate β HEIG-VD & EPFL (ESO Lab)Shanci Li
Scientific Collaborator (AI) β HEIG-VD
A new multimodal dataset "GridNet-HD" specifically designed for 3D semantic segmentation of electrical infrastructures.
-
GridNet HD Leaderboard
π₯2GridNet HD Leaderboard!
-
heig-vd-geo/GridNet-HD
Viewer β’ Updated β’ 800 β’ 3.78k β’ 15 -
GridNet-HD: A High-Resolution Multi-Modal Dataset for LiDAR-Image Fusion on Power Line Infrastructure
Paper β’ 2601.13052 β’ Published -
heig-vd-geo/PTv3_GridNet-HD_baseline
Updated β’ 1
CASWiT is a context-aware Transformer for ultra-high resolution aerial image segmentation. CASWiT achieves SOTA performances on FLAIR-HUB and URUR.
A new multimodal dataset "GridNet-HD" specifically designed for 3D semantic segmentation of electrical infrastructures.
-
GridNet HD Leaderboard
π₯2GridNet HD Leaderboard!
-
heig-vd-geo/GridNet-HD
Viewer β’ Updated β’ 800 β’ 3.78k β’ 15 -
GridNet-HD: A High-Resolution Multi-Modal Dataset for LiDAR-Image Fusion on Power Line Infrastructure
Paper β’ 2601.13052 β’ Published -
heig-vd-geo/PTv3_GridNet-HD_baseline
Updated β’ 1
models 12
heig-vd-geo/CASWiT
Image Segmentation β’ Updated β’ 7
heig-vd-geo/DeepChoice
Updated
heig-vd-geo/13-CASWiT-Hybrid-LandsatV2
Updated
heig-vd-geo/NTIRE2026-infraredSR-team13
Updated
heig-vd-geo/ISDNet-pytorch
Image Segmentation β’ Updated
heig-vd-geo/Glacial-Lakes
Updated β’ 1
heig-vd-geo/SPT_GridNet-HD_baseline
Other β’ Updated β’ 3
heig-vd-geo/ImageVote_GridNet-HD_baseline
Image Segmentation β’ Updated β’ 7 β’ 5
heig-vd-geo/LateFusionMLP_GridNet-HD_baseline
Other β’ Updated β’ 10 β’ 6
heig-vd-geo/PTv3_GridNet-HD_baseline
Updated β’ 1
datasets 10
heig-vd-geo/GridNet-HD
Viewer β’ Updated β’ 800 β’ 3.78k β’ 15
heig-vd-geo/13-CASWiT-Hybrid-LandsatV2-results
Viewer β’ Updated β’ 223 β’ 31
heig-vd-geo/3DSES
Updated β’ 804
heig-vd-geo/ImagesAndPointCloudsCulturalHeritageDataset
Viewer β’ Updated β’ 600 β’ 318
heig-vd-geo/piscine-results
Viewer β’ Updated β’ 1 β’ 22
heig-vd-geo/projet-piscine
Updated β’ 27
heig-vd-geo/URUR
Preview β’ Updated β’ 66
heig-vd-geo/FLAIR1_mmseg
Updated β’ 169 β’ 1
heig-vd-geo/M3DRS
Updated β’ 352 β’ 6
heig-vd-geo/STDL-soils
Viewer β’ Updated β’ 7.11k β’ 145 β’ 1