Hugging Face's logo

peldrak
/
segformer-b4-cityscapes-finetuned-grCoastline

Image Segmentation
Transformers
TensorBoard
Safetensors
segformer
vision
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
segformer-b4-cityscapes-finetuned-grCoastline / README.md
peldrak's picture
peldrak
End of training
9140d54 verified
preview code
|
raw
history blame contribute delete
No virus
20.3 kB
metadata
license: other
base_model: nvidia/segformer-b4-finetuned-cityscapes-1024-1024
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: segformer-b4-cityscapes-finetuned-grCoastline
    results: []

segformer-b4-cityscapes-finetuned-grCoastline

This model is a fine-tuned version of nvidia/segformer-b4-finetuned-cityscapes-1024-1024 on the peldrak/grCoastline_512 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2906
  • Mean Iou: 0.6957
  • Mean Accuracy: 0.7566
  • Overall Accuracy: 0.9338
  • Accuracy Water: 0.9885
  • Accuracy Whitewater: 0.0
  • Accuracy Sediment: 0.9414
  • Accuracy Other Natural Terrain: 0.6287
  • Accuracy Vegetation: 0.9357
  • Accuracy Development: 0.8053
  • Accuracy Unknown: 0.9966
  • Iou Water: 0.9551
  • Iou Whitewater: 0.0
  • Iou Sediment: 0.8065
  • Iou Other Natural Terrain: 0.5720
  • Iou Vegetation: 0.8112
  • Iou Development: 0.7293
  • Iou Unknown: 0.9955
  • F1 Score: 0.9309

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
1.6107 0.24 20 1.3419 0.4266 0.5344 0.8055 0.8982 0.0022 0.9217 0.0719 0.8391 0.0219 0.9860 0.8379 0.0011 0.4894 0.0656 0.5876 0.0215 0.9828 0.7711
1.1717 0.49 40 0.8217 0.4535 0.5561 0.8363 0.9603 0.0 0.9248 0.0754 0.9495 0.0003 0.9825 0.8828 0.0 0.5861 0.0723 0.6520 0.0003 0.9812 0.7925
0.8573 0.73 60 0.5997 0.4703 0.5698 0.8490 0.9803 0.0 0.9321 0.1462 0.9391 0.0002 0.9907 0.8943 0.0 0.5475 0.1397 0.7223 0.0002 0.9881 0.8114
0.7193 0.98 80 0.4702 0.5150 0.6015 0.8700 0.9545 0.0 0.9486 0.3257 0.9649 0.0215 0.9954 0.9238 0.0 0.6830 0.2889 0.6955 0.0214 0.9921 0.8439
0.5437 1.22 100 0.4813 0.5002 0.5954 0.8540 0.9882 0.0 0.9254 0.4698 0.7725 0.0147 0.9976 0.8106 0.0 0.6624 0.3822 0.6389 0.0147 0.9927 0.8316
0.8658 1.46 120 0.3638 0.5858 0.6575 0.9009 0.9810 0.0 0.9237 0.6942 0.9002 0.1056 0.9981 0.9410 0.0 0.8170 0.5000 0.7477 0.1050 0.9902 0.8879
0.5341 1.71 140 0.3070 0.6440 0.7090 0.9171 0.9735 0.0 0.9238 0.6888 0.9196 0.4608 0.9966 0.9364 0.0 0.8131 0.5558 0.7716 0.4381 0.9930 0.9131
0.8529 1.95 160 0.3170 0.6007 0.6686 0.8978 0.9662 0.0 0.8928 0.4959 0.9625 0.3657 0.9969 0.9252 0.0 0.7920 0.4283 0.7162 0.3487 0.9941 0.8893
0.8267 2.2 180 0.3040 0.6343 0.6983 0.9128 0.9786 0.0 0.9172 0.5689 0.9531 0.4730 0.9973 0.9321 0.0 0.7818 0.5206 0.7544 0.4585 0.9928 0.9066
0.5768 2.44 200 0.2663 0.6589 0.7256 0.9206 0.9715 0.0 0.9315 0.6199 0.9318 0.6269 0.9977 0.9466 0.0 0.7959 0.5364 0.7784 0.5601 0.9949 0.9173
0.5688 2.68 220 0.2594 0.6610 0.7289 0.9195 0.9628 0.0 0.9420 0.6217 0.9207 0.6572 0.9980 0.9410 0.0 0.8027 0.5390 0.7653 0.5846 0.9944 0.9166
0.5779 2.93 240 0.2871 0.6267 0.7002 0.9104 0.9725 0.0 0.9601 0.7173 0.8653 0.3899 0.9965 0.9460 0.0 0.7667 0.5370 0.7629 0.3798 0.9948 0.9064
0.4996 3.17 260 0.2841 0.6546 0.7249 0.9170 0.9716 0.0 0.9684 0.5552 0.9162 0.6657 0.9974 0.9480 0.0 0.7529 0.4863 0.7826 0.6172 0.9949 0.9129
0.4896 3.41 280 0.2394 0.6713 0.7389 0.9240 0.9816 0.0 0.9350 0.6015 0.9215 0.7352 0.9976 0.9440 0.0 0.7865 0.5442 0.7849 0.6454 0.9944 0.9206
0.26 3.66 300 0.2547 0.6699 0.7440 0.9203 0.9620 0.0 0.9679 0.7774 0.8282 0.6764 0.9960 0.9355 0.0 0.7622 0.5947 0.7708 0.6317 0.9946 0.9200
0.1991 3.9 320 0.2875 0.6528 0.7179 0.9133 0.9535 0.0 0.8864 0.5871 0.9496 0.6516 0.9969 0.9309 0.0 0.8090 0.5008 0.7420 0.5919 0.9948 0.9104
0.3001 4.15 340 0.2382 0.6728 0.7437 0.9227 0.9653 0.0 0.9621 0.7494 0.8552 0.6778 0.9963 0.9388 0.0 0.7812 0.5856 0.7809 0.6285 0.9947 0.9220
0.2449 4.39 360 0.2335 0.6835 0.7610 0.9259 0.9752 0.0 0.9335 0.8187 0.8107 0.7907 0.9982 0.9431 0.0 0.7791 0.6198 0.7808 0.6673 0.9945 0.9260
0.1596 4.63 380 0.2749 0.6713 0.7359 0.9244 0.9762 0.0 0.9506 0.5984 0.9361 0.6939 0.9958 0.9502 0.0 0.8048 0.5367 0.7810 0.6316 0.9946 0.9210
0.4593 4.88 400 0.2356 0.6888 0.7552 0.9319 0.9833 0.0 0.9431 0.7303 0.8890 0.7433 0.9975 0.9368 0.0 0.7817 0.6452 0.8107 0.6519 0.9950 0.9303
0.2765 5.12 420 0.2227 0.6872 0.7495 0.9317 0.9697 0.0 0.9500 0.6918 0.9300 0.7090 0.9962 0.9480 0.0 0.7901 0.6052 0.8154 0.6567 0.9949 0.9298
0.239 5.37 440 0.2204 0.6962 0.7653 0.9347 0.9827 0.0 0.9260 0.7312 0.8975 0.8233 0.9965 0.9534 0.0 0.7928 0.6387 0.8146 0.6790 0.9949 0.9336
0.2015 5.61 460 0.2473 0.6902 0.7586 0.9321 0.9777 0.0 0.9450 0.6361 0.9254 0.8290 0.9970 0.9559 0.0 0.7907 0.5878 0.8066 0.6947 0.9954 0.9296
0.1847 5.85 480 0.2867 0.6724 0.7363 0.9252 0.9736 0.0 0.9011 0.6840 0.9348 0.6652 0.9950 0.9362 0.0 0.7855 0.5820 0.7988 0.6102 0.9942 0.9230
0.1476 6.1 500 0.2539 0.6841 0.7534 0.9280 0.9722 0.0 0.9291 0.6154 0.9258 0.8335 0.9980 0.9508 0.0 0.7905 0.5546 0.7932 0.7040 0.9954 0.9253
0.199 6.34 520 0.2064 0.6957 0.7675 0.9336 0.9897 0.0 0.8901 0.7931 0.8683 0.8350 0.9965 0.9448 0.0 0.7757 0.6496 0.8200 0.6846 0.9949 0.9331
0.1857 6.59 540 0.3659 0.6484 0.7182 0.9115 0.9770 0.0 0.9370 0.3944 0.9597 0.7617 0.9974 0.9496 0.0 0.8004 0.3773 0.7353 0.6808 0.9955 0.9026
0.1589 6.83 560 0.2996 0.6793 0.7430 0.9268 0.9728 0.0 0.9351 0.5843 0.9490 0.7625 0.9977 0.9500 0.0 0.8034 0.5358 0.7869 0.6836 0.9954 0.9232
0.1164 7.07 580 0.2593 0.6962 0.7629 0.9339 0.9802 0.0 0.9348 0.7146 0.9014 0.8125 0.9969 0.9544 0.0 0.7944 0.6129 0.8139 0.7022 0.9954 0.9326
0.1889 7.32 600 0.2391 0.6957 0.7579 0.9353 0.9755 0.0 0.9414 0.7279 0.9161 0.7465 0.9983 0.9542 0.0 0.7972 0.6167 0.8285 0.6780 0.9954 0.9339
0.1553 7.56 620 0.2779 0.6909 0.7537 0.9322 0.9802 0.0 0.9396 0.6127 0.9428 0.8028 0.9980 0.9529 0.0 0.8069 0.5620 0.8077 0.7112 0.9956 0.9290
0.1076 7.8 640 0.2944 0.6883 0.7516 0.9302 0.9733 0.0 0.9371 0.5937 0.9519 0.8084 0.9968 0.9515 0.0 0.8022 0.5417 0.8030 0.7239 0.9959 0.9269
0.2934 8.05 660 0.2662 0.6932 0.7642 0.9328 0.9834 0.0 0.9042 0.6589 0.9175 0.8868 0.9986 0.9514 0.0 0.7865 0.5925 0.8158 0.7107 0.9953 0.9307
0.1314 8.29 680 0.2920 0.6884 0.7516 0.9314 0.9748 0.0 0.9613 0.6612 0.9234 0.7441 0.9965 0.9519 0.0 0.7869 0.5894 0.8092 0.6860 0.9955 0.9291
0.1798 8.54 700 0.2516 0.6888 0.7547 0.9302 0.9815 0.0 0.9121 0.6543 0.9264 0.8125 0.9958 0.9488 0.0 0.7871 0.5725 0.8084 0.7099 0.9948 0.9280
0.082 8.78 720 0.2831 0.6901 0.7526 0.9317 0.9787 0.0 0.9223 0.6395 0.9422 0.7885 0.9973 0.9559 0.0 0.7998 0.5720 0.8051 0.7025 0.9958 0.9291
0.4488 9.02 740 0.2693 0.6909 0.7541 0.9316 0.9830 0.0 0.9435 0.6212 0.9308 0.8024 0.9976 0.9578 0.0 0.7988 0.5632 0.8022 0.7184 0.9956 0.9287
0.157 9.27 760 0.2555 0.6938 0.7586 0.9329 0.9811 0.0 0.9349 0.6863 0.9092 0.7999 0.9986 0.9523 0.0 0.7878 0.5975 0.8151 0.7087 0.9952 0.9311
0.3359 9.51 780 0.2756 0.6931 0.7595 0.9313 0.9710 0.0 0.9391 0.6719 0.9094 0.8264 0.9990 0.9498 0.0 0.7894 0.5864 0.8044 0.7262 0.9954 0.9295
0.1014 9.76 800 0.3388 0.6731 0.7371 0.9228 0.9806 0.0 0.9065 0.5523 0.9482 0.7741 0.9980 0.9459 0.0 0.7898 0.4996 0.7776 0.7025 0.9962 0.9186
0.2797 10.0 820 0.2873 0.6859 0.7484 0.9294 0.9781 0.0 0.9256 0.6384 0.9330 0.7657 0.9979 0.9531 0.0 0.7977 0.5552 0.8002 0.6986 0.9965 0.9269
0.0689 10.24 840 0.2685 0.6996 0.7604 0.9364 0.9843 0.0 0.9467 0.6825 0.9266 0.7858 0.9972 0.9571 0.0 0.8004 0.6079 0.8236 0.7122 0.9959 0.9344
0.1475 10.49 860 0.2654 0.6986 0.7611 0.9350 0.9778 0.0 0.9298 0.6551 0.9378 0.8288 0.9982 0.9559 0.0 0.8022 0.5867 0.8177 0.7313 0.9963 0.9327
0.0768 10.73 880 0.2562 0.6916 0.7576 0.9299 0.9812 0.0 0.9241 0.7658 0.8645 0.7692 0.9982 0.9551 0.0 0.7959 0.5885 0.7989 0.7060 0.9965 0.9296
0.1461 10.98 900 0.3047 0.6927 0.7544 0.9326 0.9780 0.0006 0.9205 0.6334 0.9469 0.8030 0.9982 0.9543 0.0006 0.8019 0.5720 0.8075 0.7159 0.9965 0.9299
0.1344 11.22 920 0.2906 0.6957 0.7566 0.9338 0.9885 0.0 0.9414 0.6287 0.9357 0.8053 0.9966 0.9551 0.0 0.8065 0.5720 0.8112 0.7293 0.9955 0.9309

Framework versions

  • Transformers 4.38.1
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2