sam1120's picture
update model card README.md
ccf0d18
metadata
license: other
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: dropoff-utcustom-train-SF-RGBD-b0_6
    results: []

dropoff-utcustom-train-SF-RGBD-b0_6

This model is a fine-tuned version of nvidia/mit-b0 on the sam1120/dropoff-utcustom-TRAIN dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2353
  • Mean Iou: 0.6539
  • Mean Accuracy: 0.7065
  • Overall Accuracy: 0.9662
  • Accuracy Unlabeled: nan
  • Accuracy Dropoff: 0.4233
  • Accuracy Undropoff: 0.9897
  • Iou Unlabeled: nan
  • Iou Dropoff: 0.3423
  • Iou Undropoff: 0.9656

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: 7e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.05
  • num_epochs: 120

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Accuracy Unlabeled Accuracy Dropoff Accuracy Undropoff Iou Unlabeled Iou Dropoff Iou Undropoff
0.9975 5.0 10 1.0470 0.2819 0.6747 0.7186 nan 0.6267 0.7226 0.0 0.1290 0.7167
0.8329 10.0 20 0.8435 0.3211 0.5026 0.9526 nan 0.0117 0.9934 0.0 0.0106 0.9526
0.6857 15.0 30 0.6184 0.3191 0.4994 0.9567 nan 0.0006 0.9981 0.0 0.0006 0.9567
0.5913 20.0 40 0.4793 0.3193 0.4997 0.9573 nan 0.0005 0.9988 0.0 0.0005 0.9573
0.5299 25.0 50 0.4529 0.3488 0.5442 0.9596 nan 0.0911 0.9973 0.0 0.0869 0.9595
0.4922 30.0 60 0.4037 0.4352 0.6983 0.9671 nan 0.4051 0.9915 0.0 0.3390 0.9666
0.4769 35.0 70 0.4161 0.4090 0.7560 0.9426 nan 0.5524 0.9595 0.0 0.2858 0.9412
0.3916 40.0 80 0.3343 0.6320 0.6946 0.9614 nan 0.4036 0.9856 nan 0.3033 0.9608
0.3567 45.0 90 0.3372 0.6374 0.7140 0.9598 nan 0.4458 0.9821 nan 0.3157 0.9591
0.3234 50.0 100 0.3074 0.6402 0.6883 0.9652 nan 0.3863 0.9903 nan 0.3157 0.9646
0.3181 55.0 110 0.3043 0.6396 0.7138 0.9606 nan 0.4446 0.9830 nan 0.3194 0.9599
0.2584 60.0 120 0.3069 0.6450 0.7204 0.9613 nan 0.4576 0.9831 nan 0.3294 0.9605
0.2566 65.0 130 0.2824 0.6431 0.7063 0.9630 nan 0.4263 0.9863 nan 0.3239 0.9623
0.2353 70.0 140 0.2763 0.6470 0.7046 0.9645 nan 0.4212 0.9880 nan 0.3301 0.9638
0.2368 75.0 150 0.2644 0.6474 0.6973 0.9658 nan 0.4044 0.9902 nan 0.3296 0.9652
0.2225 80.0 160 0.2673 0.6462 0.7089 0.9635 nan 0.4313 0.9866 nan 0.3296 0.9629
0.1976 85.0 170 0.2568 0.6449 0.7057 0.9637 nan 0.4244 0.9870 nan 0.3268 0.9630
0.1981 90.0 180 0.2572 0.6444 0.7110 0.9626 nan 0.4365 0.9855 nan 0.3269 0.9619
0.1857 95.0 190 0.2503 0.6504 0.7027 0.9658 nan 0.4157 0.9897 nan 0.3356 0.9652
0.1826 100.0 200 0.2345 0.6509 0.6984 0.9666 nan 0.4059 0.9909 nan 0.3357 0.9660
0.1818 105.0 210 0.2484 0.6506 0.7160 0.9637 nan 0.4458 0.9862 nan 0.3381 0.9630
0.1919 110.0 220 0.2343 0.6526 0.6996 0.9669 nan 0.4080 0.9912 nan 0.3389 0.9663
0.17 115.0 230 0.2377 0.6535 0.7065 0.9661 nan 0.4235 0.9896 nan 0.3416 0.9655
0.1739 120.0 240 0.2353 0.6539 0.7065 0.9662 nan 0.4233 0.9897 nan 0.3423 0.9656

Framework versions

  • Transformers 4.30.2
  • Pytorch 2.0.1+cu117
  • Datasets 2.13.1
  • Tokenizers 0.13.3