Downloads · 30 days
0
geonusaf-s4-unetformer-r18-R1-block-fold1
Pranilllllll/geonusaf-s4-unetformer-r18-R1-block-fold1
geonusaf-s4-unetformer-r18-R1-block-fold1 is a machine learning model from Pranilllllll. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for pytorch.
Kathmandu Valley land-use segmentation, 6 classes, ignoreindex=255.
Access
Public
Updated Sep 3, 2026
Repo size
2.3 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt235 MB · 98%
At a glance
- Library
- pytorch
- Access
- Public
- Created
- Sep 3, 2026
- Updated
- Sep 3, 2026
- SHA
- 16787455
Model card
From the Hugging Face model README
GeoNUSAF Stage 4 - UNetFormer (ResNet-18) - arm R1 - block fold 1
Kathmandu Valley land-use segmentation, 6 classes, ignore_index=255.
| field | value |
|---|---|
| architecture | UNetFormer, timm resnet18 encoder (ImageNet), global-local attention decoder |
| arch version | unetformer-r18-v2 (sig 29554ac657c8) |
| arm | R1 (real + 804 synthetic) |
| train pairs | 1608 (804 real + 804 synthetic) |
| synthetic source | sugam24/geonusaf-stage3-fakepairs-block-fold1, fake_pairs_R1 |
| split | block fold 1, train_sha1 eb0aebdd9919 |
| input | 512x512, ImageNet norm, working GSD 1.374 m/px |
| lr (dec/enc) | 0.0003 / 3e-05, AdamW wd 0.01 |
| schedule | 6000 steps, warmup 500, cosine 6000 |
| regularization | EMA 0.999, label smoothing 0.05, drop path 0.1, dropout 0.1 |
| aux head weight | 0.4 |
| class weights | from real |
| weights | EMA |
| seed | 42 |
| best step | 5200 |
| val mIoU | 0.4897 |
| val mF1 | 0.6290 |
| val OA | 0.7975 |
| val kappa | 0.6543 |
Per-class (validation, 136 real fold-1 tiles)
| class | IoU | F1 |
|---|---|---|
| Residential | 0.8222 | 0.9024 |
| Road | 0.3357 | 0.5027 |
| River | 0.2281 | 0.3715 |
| Forest | 0.6946 | 0.8198 |
| UnusedLand | 0.3006 | 0.4623 |
| Agricultural | 0.5568 | 0.7153 |
Validation contains no synthetic pixels in either arm.
Not comparable to the part-1 fold-1 UNetFormer run: that one used an epoch ceiling, epoch-granularity early stopping and epoch-seeded augmentation.
Architecture is an independent implementation of Wang et al. (2022), ISPRS J. Photogramm. Remote Sens. 190:196-214 (the reference repo is GPL-3.0).
LR deviates from the UNetFormer paper's 6e-4 (reduced to 3e-4 for parity with the other baselines at this dataset scale).
Checkpoint best.pt holds model_state (EMA weights) plus cfg, metrics, arch_sig.
- Library
- pytorch
- Created
- Sep 3, 2026
- Updated
- Sep 3, 2026