<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
test_buckinghamshire_model
This model is a fine-tuned version of ratnaonline1/segFormer-b4-city-satellite-segmentation-1024x1024 on the rwood-97/test_buckinghamshire dataset. It achieves the following results on the evaluation set:
- Loss: 0.9454
- Mean Iou: 0.1942
- Mean Accuracy: 0.4998
- Overall Accuracy: 0.3876
- Accuracy Non-map: 0.9984
- Accuracy Map: 0.0012
- Iou Non-map: 0.3871
- Iou Map: 0.0012
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 6e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Non-map | Accuracy Map | Iou Non-map | Iou Map |
|---|---|---|---|---|---|---|---|---|---|---|
| 0.6104 | 2.22 | 20 | 0.7657 | 0.2215 | 0.4968 | 0.4014 | 0.9206 | 0.0730 | 0.3734 | 0.0695 |
| 0.7027 | 4.44 | 40 | 0.7235 | 0.2895 | 0.4947 | 0.4514 | 0.6872 | 0.3022 | 0.3268 | 0.2523 |
| 0.5333 | 6.67 | 60 | 0.7271 | 0.2734 | 0.4955 | 0.4373 | 0.7541 | 0.2369 | 0.3418 | 0.2050 |
| 0.6081 | 8.89 | 80 | 0.7233 | 0.2824 | 0.4978 | 0.4456 | 0.7296 | 0.2660 | 0.3377 | 0.2271 |
| 0.6295 | 11.11 | 100 | 0.7889 | 0.2124 | 0.4985 | 0.3969 | 0.9496 | 0.0473 | 0.3789 | 0.0459 |
| 0.5739 | 13.33 | 120 | 0.8912 | 0.1952 | 0.4996 | 0.3880 | 0.9954 | 0.0038 | 0.3866 | 0.0038 |
| 0.5729 | 15.56 | 140 | 0.9120 | 0.1940 | 0.4999 | 0.3876 | 0.9991 | 0.0007 | 0.3873 | 0.0007 |
| 0.5813 | 17.78 | 160 | 0.9248 | 0.1941 | 0.4999 | 0.3876 | 0.9988 | 0.0009 | 0.3872 | 0.0009 |
| 0.6749 | 20.0 | 180 | 0.9454 | 0.1942 | 0.4998 | 0.3876 | 0.9984 | 0.0012 | 0.3871 | 0.0012 |
Framework versions
- Transformers 4.35.0.dev0
- Pytorch 2.1.0
- Datasets 2.14.5
- Tokenizers 0.14.1
