segformer-b5-ade-finetuned-coastTrain-grCoastline
This model is a fine-tuned version of peldrak/segformer-b5-ade-finetuned-coastTrain on the peldrak/grCoastline_512 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1862
- Mean Iou: 0.7261
- Mean Accuracy: 0.7941
- Overall Accuracy: 0.9457
- Accuracy Water: 0.9892
- Accuracy Whitewater: 0.1212
- Accuracy Sediment: 0.8832
- Accuracy Other Natural Terrain: 0.9089
- Accuracy Vegetation: 0.8620
- Accuracy Development: 0.7955
- Accuracy Unknown: 0.9987
- Iou Water: 0.9551
- Iou Whitewater: 0.1179
- Iou Sediment: 0.8407
- Iou Other Natural Terrain: 0.7579
- Iou Vegetation: 0.8186
- Iou Development: 0.5970
- Iou Unknown: 0.9957
- F1 Score: 0.9457
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 Water | Accuracy Whitewater | Accuracy Sediment | Accuracy Other Natural Terrain | Accuracy Vegetation | Accuracy Development | Accuracy Unknown | Iou Water | Iou Whitewater | Iou Sediment | Iou Other Natural Terrain | Iou Vegetation | Iou Development | Iou Unknown | F1 Score |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.3949 | 0.24 | 20 | 0.4674 | 0.5533 | 0.6548 | 0.8647 | 0.9629 | 0.0 | 0.8258 | 0.3380 | 0.8635 | 0.5954 | 0.9980 | 0.8563 | 0.0 | 0.5989 | 0.2978 | 0.6772 | 0.4600 | 0.9826 | 0.8527 |
0.3836 | 0.49 | 40 | 0.3509 | 0.6135 | 0.6996 | 0.8983 | 0.9860 | 0.0 | 0.9063 | 0.4811 | 0.8910 | 0.6351 | 0.9974 | 0.9202 | 0.0 | 0.7089 | 0.4491 | 0.6990 | 0.5252 | 0.9922 | 0.8914 |
0.4461 | 0.73 | 60 | 0.2453 | 0.6595 | 0.7407 | 0.9222 | 0.9948 | 0.0 | 0.8439 | 0.8332 | 0.8203 | 0.6988 | 0.9937 | 0.9202 | 0.0 | 0.7836 | 0.6604 | 0.7620 | 0.4984 | 0.9916 | 0.9219 |
0.7886 | 0.98 | 80 | 0.2794 | 0.6594 | 0.7302 | 0.9222 | 0.9841 | 0.0004 | 0.8601 | 0.6668 | 0.9358 | 0.6682 | 0.9962 | 0.9314 | 0.0004 | 0.8039 | 0.6086 | 0.7510 | 0.5271 | 0.9931 | 0.9198 |
0.2695 | 1.22 | 100 | 0.2217 | 0.6871 | 0.7557 | 0.9326 | 0.9879 | 0.0 | 0.9150 | 0.7878 | 0.8622 | 0.7392 | 0.9978 | 0.9159 | 0.0 | 0.8153 | 0.7031 | 0.7691 | 0.6121 | 0.9946 | 0.9311 |
0.2343 | 1.46 | 120 | 0.2122 | 0.6772 | 0.7477 | 0.9312 | 0.9894 | 0.0001 | 0.9146 | 0.8691 | 0.8144 | 0.6484 | 0.9975 | 0.9315 | 0.0001 | 0.7897 | 0.7006 | 0.7760 | 0.5482 | 0.9943 | 0.9300 |
0.1973 | 1.71 | 140 | 0.1803 | 0.6962 | 0.7646 | 0.9392 | 0.9769 | 0.0052 | 0.9011 | 0.8906 | 0.8519 | 0.7277 | 0.9988 | 0.9530 | 0.0052 | 0.8240 | 0.7305 | 0.7956 | 0.5717 | 0.9932 | 0.9387 |
0.2983 | 1.95 | 160 | 0.1978 | 0.6927 | 0.7630 | 0.9372 | 0.9918 | 0.0 | 0.9197 | 0.8532 | 0.8443 | 0.7362 | 0.9955 | 0.9476 | 0.0 | 0.8177 | 0.7287 | 0.7781 | 0.5832 | 0.9936 | 0.9363 |
0.1164 | 2.2 | 180 | 0.2010 | 0.6933 | 0.7540 | 0.9368 | 0.9880 | 0.0150 | 0.8760 | 0.8708 | 0.8652 | 0.6641 | 0.9990 | 0.9438 | 0.0150 | 0.8319 | 0.7074 | 0.7900 | 0.5732 | 0.9916 | 0.9357 |
0.1998 | 2.44 | 200 | 0.1988 | 0.6920 | 0.7694 | 0.9365 | 0.9824 | 0.0186 | 0.8837 | 0.9011 | 0.8253 | 0.7768 | 0.9977 | 0.9502 | 0.0186 | 0.8128 | 0.7331 | 0.7836 | 0.5517 | 0.9941 | 0.9365 |
0.5358 | 2.68 | 220 | 0.1884 | 0.6989 | 0.7633 | 0.9392 | 0.9887 | 0.0322 | 0.8630 | 0.8721 | 0.8830 | 0.7074 | 0.9969 | 0.9521 | 0.0322 | 0.8189 | 0.7330 | 0.7929 | 0.5683 | 0.9948 | 0.9386 |
0.184 | 2.93 | 240 | 0.2723 | 0.6598 | 0.7425 | 0.9222 | 0.9888 | 0.0024 | 0.8640 | 0.6604 | 0.9037 | 0.7791 | 0.9990 | 0.9246 | 0.0024 | 0.7290 | 0.6206 | 0.7797 | 0.5689 | 0.9936 | 0.9198 |
0.1789 | 3.17 | 260 | 0.3020 | 0.6541 | 0.7355 | 0.9089 | 0.9738 | 0.0623 | 0.8362 | 0.9736 | 0.6716 | 0.6330 | 0.9981 | 0.9485 | 0.0620 | 0.7995 | 0.6046 | 0.6513 | 0.5177 | 0.9953 | 0.9099 |
0.1925 | 3.41 | 280 | 0.1866 | 0.7052 | 0.7741 | 0.9420 | 0.9869 | 0.0404 | 0.8583 | 0.8487 | 0.9004 | 0.7852 | 0.9990 | 0.9463 | 0.0404 | 0.8083 | 0.7554 | 0.8171 | 0.5749 | 0.9943 | 0.9416 |
0.1533 | 3.66 | 300 | 0.1827 | 0.7119 | 0.7673 | 0.9460 | 0.9870 | 0.0374 | 0.8885 | 0.8771 | 0.9152 | 0.6690 | 0.9968 | 0.9497 | 0.0374 | 0.8259 | 0.7684 | 0.8228 | 0.5836 | 0.9951 | 0.9448 |
0.3405 | 3.9 | 320 | 0.1840 | 0.7068 | 0.7791 | 0.9420 | 0.9918 | 0.0417 | 0.8530 | 0.9050 | 0.8582 | 0.8061 | 0.9978 | 0.9467 | 0.0416 | 0.8163 | 0.7590 | 0.8089 | 0.5793 | 0.9955 | 0.9419 |
0.1095 | 4.15 | 340 | 0.1938 | 0.7142 | 0.7842 | 0.9419 | 0.9842 | 0.1098 | 0.8741 | 0.9043 | 0.8602 | 0.7577 | 0.9989 | 0.9536 | 0.1080 | 0.8306 | 0.7499 | 0.8076 | 0.5549 | 0.9949 | 0.9420 |
0.0942 | 4.39 | 360 | 0.2193 | 0.7139 | 0.7905 | 0.9384 | 0.9785 | 0.1433 | 0.8578 | 0.8932 | 0.8522 | 0.8094 | 0.9987 | 0.9456 | 0.1394 | 0.8182 | 0.7408 | 0.7945 | 0.5640 | 0.9947 | 0.9388 |
0.151 | 4.63 | 380 | 0.1964 | 0.7189 | 0.7915 | 0.9388 | 0.9826 | 0.1692 | 0.8619 | 0.8826 | 0.8601 | 0.7854 | 0.9983 | 0.9491 | 0.1609 | 0.8171 | 0.7329 | 0.7912 | 0.5858 | 0.9950 | 0.9388 |
0.1181 | 4.88 | 400 | 0.2256 | 0.7023 | 0.7752 | 0.9370 | 0.9880 | 0.1247 | 0.8074 | 0.8731 | 0.8893 | 0.7453 | 0.9988 | 0.9505 | 0.1227 | 0.7869 | 0.7375 | 0.8042 | 0.5191 | 0.9953 | 0.9370 |
0.0898 | 5.12 | 420 | 0.2061 | 0.7193 | 0.7950 | 0.9429 | 0.9908 | 0.1206 | 0.8387 | 0.8846 | 0.8758 | 0.8559 | 0.9983 | 0.9504 | 0.1163 | 0.8109 | 0.7627 | 0.8149 | 0.5839 | 0.9961 | 0.9432 |
0.1225 | 5.37 | 440 | 0.1905 | 0.7361 | 0.8013 | 0.9460 | 0.9887 | 0.1963 | 0.8653 | 0.9102 | 0.8772 | 0.7726 | 0.9991 | 0.9502 | 0.1790 | 0.8365 | 0.7638 | 0.8197 | 0.6085 | 0.9953 | 0.9458 |
0.1501 | 5.61 | 460 | 0.1883 | 0.7325 | 0.7938 | 0.9453 | 0.9921 | 0.1697 | 0.8818 | 0.9120 | 0.8677 | 0.7350 | 0.9983 | 0.9480 | 0.1599 | 0.8421 | 0.7570 | 0.8113 | 0.6134 | 0.9960 | 0.9448 |
0.0686 | 5.85 | 480 | 0.2161 | 0.7313 | 0.8038 | 0.9429 | 0.9870 | 0.2042 | 0.8679 | 0.8909 | 0.8651 | 0.8126 | 0.9987 | 0.9538 | 0.1884 | 0.8378 | 0.7427 | 0.8065 | 0.5939 | 0.9960 | 0.9432 |
0.166 | 6.1 | 500 | 0.2229 | 0.7250 | 0.7891 | 0.9438 | 0.9887 | 0.1570 | 0.8673 | 0.8510 | 0.9086 | 0.7524 | 0.9985 | 0.9540 | 0.1485 | 0.8359 | 0.7417 | 0.8132 | 0.5857 | 0.9959 | 0.9434 |
0.1168 | 6.34 | 520 | 0.2215 | 0.7207 | 0.7956 | 0.9412 | 0.9911 | 0.1826 | 0.8410 | 0.9082 | 0.8622 | 0.7868 | 0.9976 | 0.9508 | 0.1697 | 0.8113 | 0.7493 | 0.8156 | 0.5523 | 0.9959 | 0.9416 |
0.1935 | 6.59 | 540 | 0.1862 | 0.7261 | 0.7941 | 0.9457 | 0.9892 | 0.1212 | 0.8832 | 0.9089 | 0.8620 | 0.7955 | 0.9987 | 0.9551 | 0.1179 | 0.8407 | 0.7579 | 0.8186 | 0.5970 | 0.9957 | 0.9457 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.1
- Downloads last month
- 7
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for peldrak/segformer-b5-ade-finetuned-coastTrain-grCoastline
Base model
nvidia/segformer-b5-finetuned-ade-640-640