segformer-b4-ade-finetuned-coastTrain

This model is a fine-tuned version of nvidia/segformer-b4-finetuned-ade-512-512 on the peldrak/coastTrain dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2784
  • Mean Iou: 0.7615
  • Mean Accuracy: 0.8569
  • Overall Accuracy: 0.9286
  • Accuracy Water: 0.9717
  • Accuracy Whitewater: 0.5408
  • Accuracy Sediment: 0.9245
  • Accuracy Other Natural Terrain: 0.8160
  • Accuracy Vegetation: 0.8979
  • Accuracy Development: 0.9242
  • Accuracy Unknown: 0.9232
  • Iou Water: 0.9253
  • Iou Whitewater: 0.4607
  • Iou Sediment: 0.8453
  • Iou Other Natural Terrain: 0.5582
  • Iou Vegetation: 0.8460
  • Iou Development: 0.8152
  • Iou Unknown: 0.8799
  • F1 Score: 0.9283

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
1.663 0.16 20 1.5090 0.4228 0.5280 0.7404 0.7454 0.1290 0.6594 0.0007 0.9158 0.4431 0.8027 0.7085 0.0661 0.5003 0.0006 0.5240 0.3651 0.7953 0.7435
1.4278 0.31 40 1.1477 0.4735 0.5606 0.8162 0.9166 0.0018 0.7142 0.0000 0.8720 0.5680 0.8513 0.8291 0.0018 0.5805 0.0000 0.6259 0.4328 0.8441 0.8075
1.5096 0.47 60 0.8839 0.4469 0.5277 0.7917 0.8650 0.0000 0.5808 0.0 0.9654 0.4206 0.8618 0.8144 0.0000 0.5120 0.0 0.5659 0.3769 0.8589 0.7834
1.2289 0.62 80 0.7240 0.4889 0.5772 0.8285 0.9213 0.0010 0.7241 0.0 0.8451 0.6604 0.8885 0.8586 0.0010 0.6189 0.0 0.6392 0.4241 0.8807 0.8226
0.7603 0.78 100 0.5905 0.5238 0.6004 0.8575 0.9406 0.0008 0.7086 0.0 0.9081 0.7470 0.8978 0.8633 0.0008 0.6391 0.0 0.6927 0.5817 0.8892 0.8479
1.1379 0.93 120 0.5625 0.5427 0.6168 0.8713 0.9660 0.0028 0.6806 0.0 0.8917 0.8819 0.8947 0.8537 0.0028 0.6139 0.0 0.7520 0.6952 0.8813 0.8597
0.7824 1.09 140 0.5163 0.5320 0.6100 0.8653 0.9728 0.0001 0.6368 0.0 0.8720 0.8917 0.8963 0.8734 0.0001 0.6084 0.0 0.7073 0.6440 0.8911 0.8541
0.6537 1.24 160 0.4595 0.5526 0.6278 0.8782 0.9497 0.0001 0.7542 0.0 0.9018 0.8883 0.9004 0.8791 0.0001 0.6505 0.0 0.7401 0.7045 0.8938 0.8682
0.7204 1.4 180 0.4031 0.5593 0.6379 0.8834 0.9572 0.0004 0.8028 0.0 0.8648 0.9356 0.9048 0.8913 0.0004 0.7384 0.0 0.7428 0.6471 0.8950 0.8746
0.663 1.55 200 0.4097 0.5592 0.6383 0.8813 0.9777 0.0 0.8640 0.0 0.8036 0.9289 0.8937 0.8718 0.0 0.7381 0.0 0.7349 0.6829 0.8869 0.8710
0.4566 1.71 220 0.3912 0.5598 0.6405 0.8813 0.9515 0.0011 0.8262 0.0 0.8494 0.9577 0.8973 0.8743 0.0011 0.7426 0.0 0.7405 0.6706 0.8895 0.8722
1.4951 1.86 240 0.3756 0.5566 0.6419 0.8804 0.9674 0.0006 0.8683 0.0 0.7963 0.9635 0.8971 0.8869 0.0006 0.7625 0.0 0.7352 0.6198 0.8909 0.8721
0.6232 2.02 260 0.3842 0.5650 0.6357 0.8869 0.9528 0.0014 0.7446 0.0 0.9224 0.9236 0.9049 0.8897 0.0014 0.6955 0.0 0.7485 0.7258 0.8940 0.8767
1.0104 2.17 280 0.3335 0.5791 0.6489 0.8974 0.9719 0.0028 0.8548 0.0 0.8796 0.9284 0.9049 0.8934 0.0028 0.7673 0.0 0.7825 0.7165 0.8912 0.8872
0.4107 2.33 300 0.3663 0.5642 0.6456 0.8855 0.9677 0.0023 0.8966 0.0 0.8048 0.9413 0.9065 0.8765 0.0023 0.7372 0.0 0.7532 0.6814 0.8985 0.8758
0.3112 2.48 320 0.3318 0.5833 0.6553 0.9006 0.9668 0.0132 0.8693 0.0 0.8855 0.9475 0.9047 0.9026 0.0132 0.7776 0.0 0.7936 0.6974 0.8991 0.8912
0.6679 2.64 340 0.3357 0.5840 0.6520 0.8979 0.9620 0.0109 0.8876 0.0008 0.8768 0.9071 0.9187 0.8819 0.0109 0.7585 0.0008 0.8002 0.7742 0.8611 0.8873
0.6522 2.79 360 0.3201 0.5850 0.6559 0.9015 0.9703 0.0209 0.8589 0.0010 0.8874 0.9440 0.9088 0.9037 0.0208 0.7665 0.0010 0.8052 0.7186 0.8794 0.8917
0.569 2.95 380 0.3227 0.5899 0.6592 0.9000 0.9738 0.0292 0.8709 0.0294 0.8731 0.9292 0.9086 0.8907 0.0291 0.7557 0.0294 0.8057 0.7445 0.8744 0.8902
0.5766 3.1 400 0.3537 0.5747 0.6401 0.8907 0.9663 0.0301 0.7411 0.0043 0.9260 0.9117 0.9013 0.8742 0.0300 0.7095 0.0043 0.7776 0.7304 0.8966 0.8804
1.1582 3.26 420 0.3125 0.6175 0.6767 0.9030 0.9783 0.0329 0.8699 0.1886 0.9039 0.8574 0.9059 0.8795 0.0325 0.7877 0.1879 0.8158 0.7437 0.8757 0.8948
0.4788 3.41 440 0.2963 0.6457 0.7017 0.9145 0.9791 0.0397 0.9018 0.2594 0.9025 0.9140 0.9151 0.8995 0.0394 0.8221 0.2566 0.8277 0.8050 0.8697 0.9069
0.2278 3.57 460 0.3154 0.6225 0.6920 0.9049 0.9683 0.1006 0.8834 0.1576 0.8780 0.9448 0.9116 0.9053 0.0996 0.7952 0.1573 0.8066 0.7248 0.8684 0.8983
0.4206 3.72 480 0.3235 0.5959 0.6553 0.9007 0.9666 0.0412 0.8435 0.0371 0.9305 0.8624 0.9060 0.9002 0.0411 0.7604 0.0371 0.7808 0.7811 0.8709 0.8912
0.3314 3.88 500 0.3323 0.6125 0.6802 0.9019 0.9602 0.1699 0.8257 0.0432 0.9168 0.9415 0.9039 0.9100 0.1663 0.7722 0.0432 0.7877 0.7446 0.8636 0.8949
0.8233 4.03 520 0.3092 0.6410 0.7039 0.9085 0.9714 0.1106 0.8881 0.2342 0.8985 0.9205 0.9042 0.9026 0.1083 0.7966 0.2326 0.8114 0.7720 0.8636 0.9022
0.3436 4.19 540 0.3070 0.6464 0.7064 0.9100 0.9816 0.1226 0.9199 0.2470 0.8824 0.8815 0.9098 0.8936 0.1205 0.7881 0.2408 0.8151 0.7747 0.8918 0.9039
0.3504 4.34 560 0.3084 0.6827 0.7459 0.9151 0.9749 0.2036 0.9063 0.4067 0.8906 0.9284 0.9107 0.9071 0.1971 0.8147 0.3798 0.8225 0.7894 0.8686 0.9111
0.3461 4.5 580 0.3091 0.7164 0.8006 0.9183 0.9687 0.3107 0.8787 0.6910 0.8991 0.9385 0.9177 0.9184 0.2873 0.8206 0.5133 0.8230 0.7781 0.8740 0.9166
0.9608 4.65 600 0.2973 0.6896 0.7680 0.9130 0.9772 0.2477 0.8727 0.5473 0.8849 0.9410 0.9055 0.9072 0.2363 0.8040 0.4150 0.8202 0.7792 0.8653 0.9101
0.2724 4.81 620 0.2947 0.7055 0.7860 0.9169 0.9732 0.3313 0.9060 0.5587 0.8825 0.9406 0.9096 0.9133 0.3037 0.8241 0.4199 0.8221 0.7875 0.8681 0.9149
0.2541 4.96 640 0.2897 0.7142 0.7929 0.9184 0.9748 0.3398 0.8850 0.6015 0.8874 0.9459 0.9163 0.9156 0.3142 0.8310 0.4651 0.8228 0.7766 0.8740 0.9166
1.337 5.12 660 0.2950 0.7033 0.7721 0.9169 0.9612 0.3796 0.9062 0.3990 0.8976 0.9436 0.9178 0.9159 0.3526 0.8199 0.3732 0.8226 0.7638 0.8749 0.9148
0.3685 5.27 680 0.2714 0.7369 0.8200 0.9227 0.9741 0.3947 0.8746 0.7536 0.9208 0.9100 0.9123 0.9194 0.3485 0.8370 0.5641 0.8375 0.7705 0.8815 0.9216
0.2901 5.43 700 0.2848 0.7373 0.8149 0.9230 0.9704 0.4050 0.9080 0.6911 0.9158 0.9008 0.9131 0.9206 0.3630 0.8271 0.5450 0.8393 0.7948 0.8710 0.9217
0.4242 5.58 720 0.2831 0.7343 0.8311 0.9195 0.9793 0.4392 0.8635 0.7955 0.8824 0.9400 0.9181 0.9140 0.3800 0.8157 0.5413 0.8335 0.7804 0.8752 0.9187
0.2186 5.74 740 0.2713 0.7400 0.8010 0.9250 0.9680 0.4074 0.9073 0.5905 0.9445 0.8784 0.9107 0.9197 0.3715 0.8330 0.5342 0.8387 0.8023 0.8810 0.9234
0.3729 5.89 760 0.2846 0.7528 0.8290 0.9233 0.9714 0.4906 0.8906 0.7037 0.9157 0.9274 0.9039 0.9199 0.4197 0.8272 0.6058 0.8448 0.7878 0.8641 0.9225
0.29 6.05 780 0.2979 0.7411 0.8274 0.9225 0.9746 0.4241 0.9098 0.7544 0.8959 0.9247 0.9085 0.9222 0.3765 0.8288 0.5630 0.8343 0.7957 0.8672 0.9215
0.1211 6.2 800 0.2962 0.7553 0.8266 0.9254 0.9722 0.4567 0.9034 0.7216 0.9218 0.8951 0.9153 0.9200 0.4076 0.8367 0.6068 0.8400 0.8034 0.8727 0.9242
0.2875 6.36 820 0.3040 0.7576 0.8358 0.9249 0.9705 0.5281 0.8742 0.7028 0.9110 0.9454 0.9184 0.9234 0.4444 0.8312 0.5986 0.8388 0.7919 0.8746 0.9243
0.1761 6.51 840 0.2623 0.7577 0.8422 0.9288 0.9742 0.4691 0.9182 0.7958 0.9109 0.9022 0.9249 0.9227 0.4026 0.8441 0.5856 0.8486 0.8191 0.8815 0.9279
0.2962 6.67 860 0.2828 0.7498 0.8469 0.9231 0.9651 0.5313 0.8896 0.7981 0.9125 0.9153 0.9166 0.9213 0.4305 0.8172 0.5763 0.8463 0.7849 0.8724 0.9228
0.3504 6.82 880 0.2912 0.7437 0.8384 0.9219 0.9793 0.4609 0.8613 0.8330 0.9077 0.9174 0.9094 0.9122 0.3974 0.8246 0.5504 0.8378 0.8159 0.8678 0.9211
0.2496 6.98 900 0.2838 0.7476 0.8480 0.9239 0.9729 0.4970 0.8875 0.8454 0.9128 0.9111 0.9092 0.9211 0.4346 0.8324 0.5334 0.8433 0.8005 0.8678 0.9235
1.2185 7.13 920 0.3104 0.7466 0.8454 0.9201 0.9732 0.5344 0.8610 0.8031 0.8965 0.9391 0.9105 0.9215 0.4476 0.8078 0.5605 0.8257 0.7948 0.8687 0.9197
0.1779 7.29 940 0.3212 0.7591 0.8515 0.9252 0.9751 0.5615 0.8844 0.7893 0.8949 0.9314 0.9236 0.9219 0.4660 0.8250 0.5769 0.8376 0.8097 0.8769 0.9247
0.4705 7.44 960 0.2663 0.7504 0.8429 0.9243 0.9776 0.4934 0.8899 0.8011 0.8994 0.9268 0.9118 0.9159 0.4335 0.8278 0.5520 0.8401 0.7987 0.8846 0.9237
0.2637 7.6 980 0.2561 0.7639 0.8449 0.9289 0.9717 0.5314 0.8805 0.7593 0.9309 0.9236 0.9172 0.9271 0.4443 0.8284 0.6024 0.8461 0.8074 0.8916 0.9284
0.1961 7.75 1000 0.2712 0.7486 0.8598 0.9250 0.9780 0.5721 0.8804 0.8483 0.8995 0.9282 0.9119 0.9184 0.4402 0.8297 0.5164 0.8466 0.8043 0.8848 0.9252
1.0785 7.91 1020 0.2494 0.7586 0.8472 0.9308 0.9752 0.5247 0.9059 0.7655 0.9142 0.9169 0.9280 0.9263 0.4457 0.8448 0.5380 0.8490 0.7996 0.9071 0.9304
0.1453 8.06 1040 0.2792 0.7454 0.8519 0.9254 0.9704 0.5225 0.8913 0.8028 0.8936 0.9675 0.9153 0.9283 0.4321 0.8367 0.5213 0.8397 0.7497 0.9099 0.9261
0.2332 8.22 1060 0.2774 0.7452 0.8434 0.9242 0.9771 0.5126 0.9145 0.7857 0.8916 0.9063 0.9157 0.9221 0.4303 0.8100 0.5271 0.8311 0.7858 0.9097 0.9240
0.1902 8.37 1080 0.2779 0.7382 0.8500 0.9228 0.9830 0.5081 0.8918 0.8470 0.8739 0.9259 0.9205 0.9178 0.4154 0.8383 0.4896 0.8345 0.7778 0.8937 0.9230
0.2892 8.53 1100 0.2735 0.7535 0.8527 0.9275 0.9726 0.5405 0.8697 0.8180 0.9147 0.9236 0.9300 0.9293 0.4539 0.8253 0.5226 0.8400 0.8145 0.8886 0.9273
0.2251 8.68 1120 0.2626 0.7627 0.8536 0.9295 0.9781 0.5566 0.9106 0.7753 0.8922 0.9349 0.9273 0.9239 0.4522 0.8465 0.5699 0.8455 0.8062 0.8949 0.9291
0.1345 8.84 1140 0.2558 0.7676 0.8542 0.9296 0.9784 0.5649 0.9180 0.7797 0.9048 0.9185 0.9151 0.9205 0.4585 0.8451 0.5930 0.8489 0.8174 0.8900 0.9292
0.18 8.99 1160 0.2737 0.7628 0.8566 0.9288 0.9781 0.5295 0.9059 0.8418 0.9066 0.9190 0.9150 0.9207 0.4404 0.8533 0.5818 0.8542 0.8169 0.8726 0.9283
0.282 9.15 1180 0.2705 0.7734 0.8487 0.9310 0.9717 0.5354 0.9220 0.7588 0.9230 0.9135 0.9165 0.9265 0.4527 0.8522 0.6342 0.8563 0.8178 0.8736 0.9304
0.1688 9.3 1200 0.2770 0.7695 0.8663 0.9317 0.9705 0.5896 0.9098 0.8282 0.9125 0.9241 0.9295 0.9315 0.4659 0.8490 0.5848 0.8541 0.8173 0.8840 0.9316
0.1043 9.46 1220 0.2784 0.7615 0.8569 0.9286 0.9717 0.5408 0.9245 0.8160 0.8979 0.9242 0.9232 0.9253 0.4607 0.8453 0.5582 0.8460 0.8152 0.8799 0.9283

Framework versions

  • Transformers 4.38.1
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2
Downloads last month
20
Safetensors
Model size
64M params
Tensor type
F32
·
Inference Examples
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-b4-ade-finetuned-coastTrain

Finetuned
(4)
this model
Finetunes
1 model