Hugging Face's logo

sungile
/
custom-object-test7

Image Segmentation
Transformers
Safetensors
segformer
vision
Generated from Trainer
Inference Endpoints
Model card Files Community
custom-object-test7 / README.md
sungile's picture
sungile
End of training
953122c verified
preview code
|
raw
history blame contribute delete
12.3 kB
metadata
library_name: transformers
license: other
base_model: nvidia/mit-b0
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: custom-object-test7
    results: []

custom-object-test7

This model is a fine-tuned version of nvidia/mit-b0 on the sungile/custom-object-masking5 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1507
  • Mean Iou: 0.3557
  • Mean Accuracy: 0.7113
  • Overall Accuracy: 0.7113
  • Accuracy Unknown: nan
  • Accuracy Background: 0.7113
  • Accuracy Object: nan
  • Iou Unknown: 0.0
  • Iou Background: 0.7113
  • Iou Object: nan

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: 2
  • eval_batch_size: 2
  • seed: 42
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 13

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Accuracy Unknown Accuracy Background Accuracy Object Iou Unknown Iou Background Iou Object
0.7249 0.25 20 0.9434 0.2871 0.8614 0.8614 nan 0.8614 nan 0.0 0.8614 0.0
0.7121 0.5 40 0.6610 0.3810 0.7620 0.7620 nan 0.7620 nan 0.0 0.7620 nan
0.4732 0.75 60 0.5195 0.3888 0.7775 0.7775 nan 0.7775 nan 0.0 0.7775 nan
0.5129 1.0 80 0.4134 0.2955 0.5910 0.5910 nan 0.5910 nan 0.0 0.5910 nan
0.4451 1.25 100 0.3473 0.2485 0.4971 0.4971 nan 0.4971 nan 0.0 0.4971 nan
0.4109 1.5 120 0.3279 0.2516 0.5033 0.5033 nan 0.5033 nan 0.0 0.5033 nan
0.3659 1.75 140 0.3014 0.2695 0.5391 0.5391 nan 0.5391 nan 0.0 0.5391 nan
0.2206 2.0 160 0.2773 0.2581 0.5163 0.5163 nan 0.5163 nan 0.0 0.5163 nan
0.2452 2.25 180 0.2575 0.2982 0.5964 0.5964 nan 0.5964 nan 0.0 0.5964 nan
0.2496 2.5 200 0.2523 0.3345 0.6690 0.6690 nan 0.6690 nan 0.0 0.6690 nan
0.1633 2.75 220 0.3160 0.4074 0.8149 0.8149 nan 0.8149 nan 0.0 0.8149 nan
0.1426 3.0 240 0.2242 0.3451 0.6903 0.6903 nan 0.6903 nan 0.0 0.6903 nan
0.1363 3.25 260 0.2225 0.3505 0.7010 0.7010 nan 0.7010 nan 0.0 0.7010 nan
0.1337 3.5 280 0.2229 0.3799 0.7599 0.7599 nan 0.7599 nan 0.0 0.7599 nan
0.1611 3.75 300 0.1971 0.3535 0.7070 0.7070 nan 0.7070 nan 0.0 0.7070 nan
0.1376 4.0 320 0.1835 0.3504 0.7007 0.7007 nan 0.7007 nan 0.0 0.7007 nan
0.1367 4.25 340 0.1735 0.3226 0.6453 0.6453 nan 0.6453 nan 0.0 0.6453 nan
0.1452 4.5 360 0.1689 0.3096 0.6192 0.6192 nan 0.6192 nan 0.0 0.6192 nan
0.1323 4.75 380 0.1741 0.3343 0.6685 0.6685 nan 0.6685 nan 0.0 0.6685 nan
0.1519 5.0 400 0.1647 0.3433 0.6866 0.6866 nan 0.6866 nan 0.0 0.6866 nan
0.1013 5.25 420 0.1645 0.3575 0.7149 0.7149 nan 0.7149 nan 0.0 0.7149 nan
0.0967 5.5 440 0.1645 0.3620 0.7241 0.7241 nan 0.7241 nan 0.0 0.7241 nan
0.1306 5.75 460 0.1646 0.3262 0.6523 0.6523 nan 0.6523 nan 0.0 0.6523 nan
0.2066 6.0 480 0.1600 0.3326 0.6651 0.6651 nan 0.6651 nan 0.0 0.6651 nan
0.0671 6.25 500 0.1546 0.3433 0.6867 0.6867 nan 0.6867 nan 0.0 0.6867 nan
0.0644 6.5 520 0.1612 0.3284 0.6568 0.6568 nan 0.6568 nan 0.0 0.6568 nan
0.0518 6.75 540 0.1575 0.3633 0.7266 0.7266 nan 0.7266 nan 0.0 0.7266 nan
0.086 7.0 560 0.1535 0.3490 0.6980 0.6980 nan 0.6980 nan 0.0 0.6980 nan
0.0602 7.25 580 0.1624 0.2988 0.5976 0.5976 nan 0.5976 nan 0.0 0.5976 nan
0.1332 7.5 600 0.1530 0.3532 0.7063 0.7063 nan 0.7063 nan 0.0 0.7063 nan
0.0494 7.75 620 0.1461 0.3732 0.7465 0.7465 nan 0.7465 nan 0.0 0.7465 nan
0.0685 8.0 640 0.1554 0.3198 0.6396 0.6396 nan 0.6396 nan 0.0 0.6396 nan
0.0492 8.25 660 0.1484 0.3563 0.7125 0.7125 nan 0.7125 nan 0.0 0.7125 nan
0.0525 8.5 680 0.1485 0.3341 0.6681 0.6681 nan 0.6681 nan 0.0 0.6681 nan
0.0911 8.75 700 0.1553 0.3257 0.6515 0.6515 nan 0.6515 nan 0.0 0.6515 nan
0.0493 9.0 720 0.1481 0.3598 0.7197 0.7197 nan 0.7197 nan 0.0 0.7197 nan
0.0445 9.25 740 0.1540 0.3536 0.7073 0.7073 nan 0.7073 nan 0.0 0.7073 nan
0.0723 9.5 760 0.1481 0.3461 0.6921 0.6921 nan 0.6921 nan 0.0 0.6921 nan
0.047 9.75 780 0.1479 0.3495 0.6990 0.6990 nan 0.6990 nan 0.0 0.6990 nan
0.0703 10.0 800 0.1489 0.3553 0.7106 0.7106 nan 0.7106 nan 0.0 0.7106 nan
0.0532 10.25 820 0.1475 0.3635 0.7270 0.7270 nan 0.7270 nan 0.0 0.7270 nan
0.0407 10.5 840 0.1481 0.3540 0.7079 0.7079 nan 0.7079 nan 0.0 0.7079 nan
0.0395 10.75 860 0.1545 0.3430 0.6861 0.6861 nan 0.6861 nan 0.0 0.6861 nan
0.1379 11.0 880 0.1531 0.3427 0.6854 0.6854 nan 0.6854 nan 0.0 0.6854 nan
0.0281 11.25 900 0.1483 0.3466 0.6932 0.6932 nan 0.6932 nan 0.0 0.6932 nan
0.0361 11.5 920 0.1560 0.3424 0.6848 0.6848 nan 0.6848 nan 0.0 0.6848 nan
0.0422 11.75 940 0.1494 0.3576 0.7152 0.7152 nan 0.7152 nan 0.0 0.7152 nan
0.0541 12.0 960 0.1502 0.3552 0.7105 0.7105 nan 0.7105 nan 0.0 0.7105 nan
0.0568 12.25 980 0.1549 0.3442 0.6885 0.6885 nan 0.6885 nan 0.0 0.6885 nan
0.0881 12.5 1000 0.1516 0.3481 0.6963 0.6963 nan 0.6963 nan 0.0 0.6963 nan
0.0478 12.75 1020 0.1527 0.3524 0.7047 0.7047 nan 0.7047 nan 0.0 0.7047 nan
0.0529 13.0 1040 0.1507 0.3557 0.7113 0.7113 nan 0.7113 nan 0.0 0.7113 nan

Framework versions

  • Transformers 4.47.1
  • Pytorch 2.1.0+cu118
  • Datasets 3.2.0
  • Tokenizers 0.21.0