Hugging Face's logo

mouadenna
/
segformer-b1-finetuned-segments-pv_v1_normalized_p100_4batch_try1

Image Segmentation
Transformers
TensorBoard
Safetensors
segformer
vision
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
segformer-b1-finetuned-segments-pv_v1_normalized_p100_4batch_try1 / README.md
mouadenna's picture
mouadenna
End of training
c6018cc verified
preview code
|
raw
history blame contribute delete
4.95 kB
metadata
license: other
base_model: nvidia/segformer-b1-finetuned-ade-512-512
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
metrics:
  - precision
model-index:
  - name: segformer-b1-finetuned-segments-pv_v1_normalized_p100_4batch_try1
    results: []

Visualize in Weights & Biases

segformer-b1-finetuned-segments-pv_v1_normalized_p100_4batch_try1

This model is a fine-tuned version of nvidia/segformer-b1-finetuned-ade-512-512 on the mouadenna/satellite_PV_dataset_train_test_v1 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0012
  • Mean Iou: 0.9586
  • Precision: 0.9792

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: 0.0004
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 40
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Mean Iou Precision
0.2548 0.9989 229 0.0851 0.6627 0.7444
0.0259 1.9978 458 0.0141 0.8187 0.8803
0.011 2.9967 687 0.0082 0.8288 0.8937
0.0073 4.0 917 0.0055 0.8596 0.8955
0.0059 4.9989 1146 0.0053 0.8527 0.8786
0.0047 5.9978 1375 0.0039 0.8920 0.9370
0.0039 6.9967 1604 0.0039 0.8811 0.9470
0.0041 8.0 1834 0.0046 0.8564 0.9432
0.0042 8.9989 2063 0.0040 0.8786 0.9099
0.004 9.9978 2292 0.0029 0.9062 0.9479
0.0037 10.9967 2521 0.0030 0.9002 0.9557
0.0031 12.0 2751 0.0026 0.9150 0.9415
0.0028 12.9989 2980 0.0023 0.9216 0.9597
0.0035 13.9978 3209 0.0038 0.8824 0.9091
0.0032 14.9967 3438 0.0029 0.9041 0.9477
0.0032 16.0 3668 0.0024 0.9191 0.9548
0.0026 16.9989 3897 0.0025 0.9177 0.9487
0.0024 17.9978 4126 0.0022 0.9235 0.9523
0.0025 18.9967 4355 0.0021 0.9270 0.9563
0.003 20.0 4585 0.0034 0.8911 0.9511
0.0027 20.9989 4814 0.0023 0.9216 0.9576
0.0024 21.9978 5043 0.0020 0.9296 0.9606
0.0023 22.9967 5272 0.0019 0.9331 0.9602
0.002 24.0 5502 0.0020 0.9318 0.9667
0.002 24.9989 5731 0.0018 0.9373 0.9619
0.0022 25.9978 5960 0.0019 0.9352 0.9582
0.0025 26.9967 6189 0.0019 0.9328 0.9686
0.0019 28.0 6419 0.0017 0.9400 0.9632
0.0018 28.9989 6648 0.0016 0.9430 0.9689
0.0017 29.9978 6877 0.0016 0.9443 0.9712
0.0017 30.9967 7106 0.0015 0.9471 0.9720
0.0016 32.0 7336 0.0015 0.9492 0.9719
0.0016 32.9989 7565 0.0014 0.9503 0.9721
0.0015 33.9978 7794 0.0014 0.9525 0.9737
0.0015 34.9967 8023 0.0013 0.9532 0.9713
0.0014 36.0 8253 0.0013 0.9536 0.9687
0.0014 36.9989 8482 0.0012 0.9562 0.9733
0.0014 37.9978 8711 0.0012 0.9576 0.9767
0.0014 38.9967 8940 0.0012 0.9579 0.9749
0.0014 39.9564 9160 0.0012 0.9586 0.9792

Framework versions

  • Transformers 4.42.3
  • Pytorch 2.1.2
  • Datasets 2.20.0
  • Tokenizers 0.19.1