CoolFace
Modelpublic

qubvel-hf/finetune-instance-segmentation-ade20k-mini-mask2former-v1

sourceHugging Faceotherupdated 2y agoView on Hugging Face
0likes15downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>

finetune-instance-segmentation-ade20k-mini-mask2former-v1

This model is a fine-tuned version of facebook/mask2former-swin-tiny-coco-instance on the qubvel-hf/ade20k-mini dataset. It achieves the following results on the evaluation set:

  • —Loss: 27.5494
  • —Map: 0.2315
  • —Map 50: 0.4495
  • —Map 75: 0.2185
  • —Map Small: 0.1535
  • —Map Medium: 0.6606
  • —Map Large: 0.8161
  • —Mar 1: 0.0981
  • —Mar 10: 0.2576
  • —Mar 100: 0.3
  • —Mar Small: 0.2272
  • —Mar Medium: 0.7189
  • —Mar Large: 0.8618
  • —Map Person: 0.1626
  • —Mar 100 Person: 0.2224
  • —Map Car: 0.3003
  • —Mar 100 Car: 0.3776

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: 1e-05
  • —trainbatchsize: 8
  • —evalbatchsize: 8
  • —seed: 42
  • —gradientaccumulationsteps: 2
  • —totaltrainbatch_size: 16
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: constant
  • —num_epochs: 40.0
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossMapMap 50Map 75Map SmallMap MediumMap LargeMar 1Mar 10Mar 100Mar SmallMar MediumMar LargeMap PersonMar 100 PersonMap CarMar 100 Car
36.78311.010033.27680.18380.36770.1740.11750.60120.79740.08840.24310.2840.21040.70530.87120.11750.20140.250.3665
30.23242.020030.82680.1980.40070.18310.13210.61830.80280.09160.250.28850.21510.71250.83540.13310.20790.2630.3691
28.41363.030029.82610.20360.4160.18490.13370.63320.79690.09340.24720.29050.21690.71620.83230.13810.21120.2690.3697
27.56594.040029.29260.21010.41760.19180.13710.63520.80510.0940.250.28840.21430.71740.83540.14560.21070.27450.3661
26.99715.050028.80440.2130.42090.20160.13790.64190.80940.0930.24990.28940.21480.72070.84410.14750.20960.27850.3692
26.426.060028.48480.21960.42240.20620.14260.6470.80460.09440.25230.29250.21880.71960.83540.150.21060.28920.3745
25.90657.070028.26010.22120.42610.2070.14440.64420.80490.09430.25270.29020.21760.71030.83230.1530.21020.28930.3703
25.67668.080028.25810.22090.42760.20760.14340.64850.82010.09430.25320.2940.21970.72120.86810.15320.21220.28850.3758
25.31119.090027.86230.22340.43180.21630.14510.6490.82520.09510.25190.29530.22120.7210.86490.15610.21480.29070.3757
24.942410.0100027.89250.22560.43670.21290.14790.64760.83140.09530.25560.29730.22440.71590.87120.15880.21530.29230.3793
24.650211.0110027.75240.22540.4410.21630.14860.64680.81860.09520.25560.29630.22310.71670.86810.15780.21530.29290.3772
24.527812.0120027.71220.22520.43490.21670.14730.64620.82370.09270.25490.29790.22510.71620.86490.15830.21650.29210.3793
24.351413.0130027.53820.2240.43450.21560.14590.65540.83240.09580.25540.29880.22510.7220.88060.15830.21910.28970.3785
24.342214.0140027.56650.2260.43740.21720.14880.65050.80590.09740.25510.29640.22410.71410.84340.15920.21580.29280.377
23.976815.0150027.77700.22810.43790.22150.14990.65530.81880.0960.25530.29780.22440.720.86320.15990.21630.29630.3793
23.700516.0160027.55350.2270.43920.21670.14850.65090.81650.09650.2550.29720.22410.71750.86560.16080.21640.29320.3779
23.57917.0170027.48940.22860.440.22090.15110.64880.81520.0970.25830.29650.22430.71130.86010.1620.21440.29520.3785
23.500418.0180027.21880.22740.43740.2160.14980.65120.79540.09620.25620.29690.22510.7120.83230.16140.2150.29330.3788
23.174419.0190027.35230.22860.43910.21660.14940.65590.82030.09620.25650.29980.22740.71560.86560.16020.21740.2970.3821
23.188420.0200027.11850.23040.43950.22040.15210.65330.80040.09680.25580.2990.22730.71310.83470.16110.2170.29980.3809
22.913621.0210027.42960.23010.43860.21970.15180.65450.81850.09680.25520.29790.22560.71230.87120.16090.21790.29920.3778
22.686322.0220026.99780.23090.4440.21960.15190.6570.79550.09760.25430.29820.22640.7140.83160.16240.21810.29940.3784
22.774123.0230027.07030.230.44360.21830.15190.65080.80290.09660.25620.30010.2290.71060.84340.1620.2180.29790.3823
22.477924.0240027.03940.23350.45210.22520.15520.6560.83180.09620.25980.30260.2310.71430.86010.16240.21870.30450.3865
22.35725.0250027.14830.23040.44560.21890.15170.65860.80650.09670.25540.29960.22780.71430.83780.1620.21870.29890.3805
22.316726.0260027.32990.2320.44380.21930.15340.65720.82210.09770.25640.29890.22670.71340.86810.16240.21760.30160.3802
22.095827.0270027.25710.2320.44380.21710.15350.65390.82680.09740.25910.29860.2260.71530.87740.16220.21850.30180.3788
22.090228.0280027.51560.23150.44820.21770.15390.65660.82650.09780.25830.30210.230.7160.87190.16260.220.30040.3842
21.994329.0290027.01420.22880.44490.21550.15110.65360.81760.0970.25570.29840.22570.71690.85690.16160.22020.29610.3766
21.884330.0300027.17380.23140.44560.21920.15340.65570.82630.09730.25870.30260.230.72040.86250.16290.22030.29990.3848
21.863531.0310027.06580.23160.44610.220.15340.65820.81660.09870.25810.30130.22920.71560.86250.1630.21880.30030.3838
21.47332.0320027.13540.23230.44930.2190.15450.65690.80770.09660.2590.30240.23050.71720.85070.16190.21820.30260.3866
21.687933.0330026.98100.23060.44610.21780.15330.65720.80950.09830.25810.30040.22850.71460.84760.16240.21940.29890.3814
21.377134.0340027.53230.230.44760.21490.15360.65930.81850.09680.25770.29960.22650.72040.86180.1620.22120.2980.3781
21.277235.0350027.14510.23270.44650.21720.15440.66410.81950.09880.25970.30280.22940.72620.85940.16160.2210.30380.3847
21.368236.0360027.46980.23340.45030.21840.1550.66080.80880.09850.25740.30130.22920.71640.85940.16570.2230.30110.3797
21.041737.0370027.24990.23540.45230.22110.15690.66430.82240.09980.26040.30370.23070.72430.85620.16540.22090.30540.3865
21.066438.0380027.34260.23040.44370.21590.15160.65680.80710.09860.25660.29930.2270.71640.84510.16410.21980.29670.3788
21.004239.0390027.77200.23150.44490.21820.15280.66110.82140.09940.25940.29940.22650.71910.85940.16040.21610.30260.3827
20.854840.0400027.54940.23150.44950.21850.15350.66060.81610.09810.25760.30.22720.71890.86180.16260.22240.30030.3776

Framework versions

  • —Transformers 4.42.0.dev0
  • —Pytorch 1.13.0+cu117
  • —Datasets 2.18.0
  • —Tokenizers 0.19.1