segformer-b4-cityscapes-finetuned-coastTrain-grCoastline
This model is a fine-tuned version of peldrak/segformer-b4-cityscapes-finetuned-coastTrain on the peldrak/grCoastline_512 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2751
- Mean Iou: 0.7477
- Mean Accuracy: 0.8306
- Overall Accuracy: 0.9274
- Accuracy Water: 0.9618
- Accuracy Whitewater: 0.5458
- Accuracy Sediment: 0.9189
- Accuracy Other Natural Terrain: 0.6111
- Accuracy Vegetation: 0.9584
- Accuracy Development: 0.8193
- Accuracy Unknown: 0.9989
- Iou Water: 0.9317
- Iou Whitewater: 0.4264
- Iou Sediment: 0.7904
- Iou Other Natural Terrain: 0.5618
- Iou Vegetation: 0.8457
- Iou Development: 0.6820
- Iou Unknown: 0.9958
- F1 Score: 0.9247
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: 30
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 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.2133 | 0.24 | 20 | 0.5121 | 0.5418 | 0.6551 | 0.8537 | 0.9585 | 0.0 | 0.8618 | 0.0673 | 0.9007 | 0.7998 | 0.9976 | 0.9023 | 0.0 | 0.6459 | 0.0662 | 0.7036 | 0.4840 | 0.9902 | 0.8244 |
0.414 | 0.49 | 40 | 0.3762 | 0.6200 | 0.7083 | 0.8876 | 0.9715 | 0.0 | 0.8827 | 0.6187 | 0.8259 | 0.6633 | 0.9959 | 0.9246 | 0.0 | 0.7128 | 0.4398 | 0.7463 | 0.5240 | 0.9926 | 0.8868 |
0.1269 | 0.73 | 60 | 0.4382 | 0.5881 | 0.6925 | 0.8750 | 0.9318 | 0.0009 | 0.8038 | 0.2996 | 0.9384 | 0.8817 | 0.9916 | 0.8967 | 0.0009 | 0.6716 | 0.2920 | 0.7370 | 0.5281 | 0.9905 | 0.8642 |
0.1997 | 0.98 | 80 | 0.3072 | 0.6505 | 0.7324 | 0.9091 | 0.9761 | 0.0012 | 0.8633 | 0.5977 | 0.9144 | 0.7777 | 0.9963 | 0.9201 | 0.0012 | 0.7241 | 0.5213 | 0.8180 | 0.5755 | 0.9932 | 0.9063 |
0.1083 | 1.22 | 100 | 0.3962 | 0.6172 | 0.6949 | 0.8968 | 0.9637 | 0.0008 | 0.9391 | 0.4493 | 0.9514 | 0.5660 | 0.9938 | 0.9245 | 0.0008 | 0.6964 | 0.4157 | 0.7968 | 0.4937 | 0.9928 | 0.8887 |
0.3105 | 1.46 | 120 | 0.2904 | 0.6512 | 0.7371 | 0.8982 | 0.9807 | 0.0948 | 0.9006 | 0.7090 | 0.8154 | 0.6621 | 0.9971 | 0.9235 | 0.0936 | 0.7400 | 0.4954 | 0.7695 | 0.5415 | 0.9947 | 0.8983 |
0.2026 | 1.71 | 140 | 0.3171 | 0.6590 | 0.7377 | 0.9117 | 0.9627 | 0.0314 | 0.9322 | 0.5886 | 0.9180 | 0.7359 | 0.9952 | 0.9374 | 0.0308 | 0.7378 | 0.5309 | 0.8117 | 0.5706 | 0.9935 | 0.9089 |
0.1502 | 1.95 | 160 | 0.2682 | 0.6830 | 0.7555 | 0.9097 | 0.9747 | 0.2278 | 0.8771 | 0.5280 | 0.9490 | 0.7334 | 0.9984 | 0.9210 | 0.2170 | 0.7785 | 0.4686 | 0.7951 | 0.6075 | 0.9934 | 0.9048 |
0.0935 | 2.2 | 180 | 0.3351 | 0.6378 | 0.7209 | 0.9009 | 0.9776 | 0.0421 | 0.9416 | 0.7188 | 0.8390 | 0.5301 | 0.9967 | 0.9143 | 0.0421 | 0.6865 | 0.5566 | 0.8033 | 0.4677 | 0.9940 | 0.8991 |
0.2849 | 2.44 | 200 | 0.2994 | 0.6750 | 0.7469 | 0.9177 | 0.9675 | 0.0987 | 0.9478 | 0.5966 | 0.9442 | 0.6781 | 0.9954 | 0.9286 | 0.0931 | 0.7319 | 0.5570 | 0.8430 | 0.5768 | 0.9943 | 0.9142 |
0.4015 | 2.68 | 220 | 0.3270 | 0.6682 | 0.7424 | 0.9058 | 0.9778 | 0.1729 | 0.8825 | 0.4336 | 0.9693 | 0.7683 | 0.9924 | 0.9117 | 0.1667 | 0.7556 | 0.4112 | 0.7914 | 0.6490 | 0.9917 | 0.8978 |
0.236 | 2.93 | 240 | 0.3329 | 0.6715 | 0.7520 | 0.9093 | 0.9707 | 0.1421 | 0.9684 | 0.6999 | 0.8499 | 0.6351 | 0.9981 | 0.9231 | 0.1363 | 0.7081 | 0.5718 | 0.8119 | 0.5550 | 0.9941 | 0.9083 |
0.0648 | 3.17 | 260 | 0.2652 | 0.6981 | 0.7843 | 0.9186 | 0.9715 | 0.1933 | 0.8879 | 0.7052 | 0.8762 | 0.8578 | 0.9983 | 0.9202 | 0.1728 | 0.7617 | 0.5836 | 0.8319 | 0.6221 | 0.9942 | 0.9184 |
0.1835 | 3.41 | 280 | 0.2945 | 0.6778 | 0.7568 | 0.9165 | 0.9734 | 0.0494 | 0.9337 | 0.6959 | 0.8648 | 0.7820 | 0.9986 | 0.9185 | 0.0483 | 0.7555 | 0.5665 | 0.8205 | 0.6412 | 0.9942 | 0.9154 |
0.0845 | 3.66 | 300 | 0.3184 | 0.6904 | 0.7687 | 0.9124 | 0.9741 | 0.2469 | 0.9595 | 0.6890 | 0.8701 | 0.6442 | 0.9969 | 0.9117 | 0.2429 | 0.7297 | 0.5577 | 0.8332 | 0.5628 | 0.9949 | 0.9112 |
0.1369 | 3.9 | 320 | 0.2729 | 0.7262 | 0.8066 | 0.9229 | 0.9706 | 0.3939 | 0.9080 | 0.5670 | 0.9483 | 0.8607 | 0.9976 | 0.9268 | 0.3526 | 0.7736 | 0.5358 | 0.8440 | 0.6550 | 0.9954 | 0.9195 |
0.2455 | 4.15 | 340 | 0.2923 | 0.7087 | 0.7941 | 0.9148 | 0.9687 | 0.3314 | 0.9364 | 0.7874 | 0.8373 | 0.6999 | 0.9978 | 0.9457 | 0.2815 | 0.7742 | 0.5592 | 0.8056 | 0.5992 | 0.9954 | 0.9160 |
0.1555 | 4.39 | 360 | 0.2976 | 0.6953 | 0.7665 | 0.9227 | 0.9694 | 0.1557 | 0.9383 | 0.6031 | 0.9438 | 0.7572 | 0.9977 | 0.9260 | 0.1480 | 0.7761 | 0.5530 | 0.8467 | 0.6235 | 0.9939 | 0.9194 |
0.1789 | 4.63 | 380 | 0.2856 | 0.7085 | 0.7802 | 0.9229 | 0.9717 | 0.2405 | 0.9347 | 0.6306 | 0.9315 | 0.7538 | 0.9983 | 0.9227 | 0.2238 | 0.7697 | 0.5606 | 0.8466 | 0.6410 | 0.9948 | 0.9202 |
0.1023 | 4.88 | 400 | 0.2929 | 0.7024 | 0.7723 | 0.9217 | 0.9639 | 0.2536 | 0.9496 | 0.6680 | 0.9365 | 0.6356 | 0.9988 | 0.9294 | 0.2279 | 0.7536 | 0.5796 | 0.8489 | 0.5825 | 0.9951 | 0.9191 |
0.1112 | 5.12 | 420 | 0.3065 | 0.7011 | 0.7720 | 0.9210 | 0.9812 | 0.1859 | 0.9282 | 0.6461 | 0.9172 | 0.7497 | 0.9955 | 0.9220 | 0.1800 | 0.7803 | 0.5507 | 0.8348 | 0.6454 | 0.9944 | 0.9187 |
0.1932 | 5.37 | 440 | 0.3091 | 0.7137 | 0.7938 | 0.9183 | 0.9588 | 0.3565 | 0.9580 | 0.6726 | 0.8990 | 0.7139 | 0.9980 | 0.9241 | 0.3106 | 0.7708 | 0.5589 | 0.8277 | 0.6077 | 0.9960 | 0.9170 |
0.092 | 5.61 | 460 | 0.3078 | 0.7082 | 0.7831 | 0.9188 | 0.9659 | 0.2639 | 0.9494 | 0.6480 | 0.9015 | 0.7550 | 0.9979 | 0.9312 | 0.2488 | 0.7690 | 0.5543 | 0.8162 | 0.6422 | 0.9957 | 0.9170 |
0.0515 | 5.85 | 480 | 0.3024 | 0.7363 | 0.8322 | 0.9208 | 0.9660 | 0.5072 | 0.9229 | 0.7492 | 0.8596 | 0.8222 | 0.9980 | 0.9318 | 0.3939 | 0.7716 | 0.5899 | 0.8228 | 0.6482 | 0.9959 | 0.9213 |
0.1004 | 6.1 | 500 | 0.3174 | 0.7201 | 0.7917 | 0.9265 | 0.9707 | 0.2714 | 0.9527 | 0.5768 | 0.9471 | 0.8254 | 0.9976 | 0.9234 | 0.2468 | 0.7821 | 0.5443 | 0.8508 | 0.6974 | 0.9958 | 0.9228 |
0.0747 | 6.34 | 520 | 0.3035 | 0.7158 | 0.7896 | 0.9233 | 0.9645 | 0.3336 | 0.9640 | 0.6184 | 0.9394 | 0.7088 | 0.9987 | 0.9279 | 0.2958 | 0.7659 | 0.5631 | 0.8517 | 0.6106 | 0.9958 | 0.9204 |
0.205 | 6.59 | 540 | 0.2899 | 0.7245 | 0.8015 | 0.9234 | 0.9788 | 0.3345 | 0.9541 | 0.7218 | 0.8790 | 0.7447 | 0.9977 | 0.9340 | 0.3038 | 0.7685 | 0.5897 | 0.8380 | 0.6410 | 0.9961 | 0.9228 |
0.0729 | 6.83 | 560 | 0.2751 | 0.7477 | 0.8306 | 0.9274 | 0.9618 | 0.5458 | 0.9189 | 0.6111 | 0.9584 | 0.8193 | 0.9989 | 0.9317 | 0.4264 | 0.7904 | 0.5618 | 0.8457 | 0.6820 | 0.9958 | 0.9247 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 3