SegFormer_Clean_Set1_95images_mit-b5_RGB

This model is a fine-tuned version of nvidia/mit-b5 on the Hasano20/Clean_Set1_95images dataset. It achieves the following results on the evaluation set:

• Loss: 0.0210
• Mean Iou: 0.9721
• Mean Accuracy: 0.9816
• Overall Accuracy: 0.9941
• Accuracy Background: 0.9974
• Accuracy Melt: 0.9506
• Accuracy Substrate: 0.9969
• Iou Background: 0.9954
• Iou Melt: 0.9316
• Iou Substrate: 0.9891

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: 0.0001
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Melt	Accuracy Substrate	Iou Background	Iou Melt	Iou Substrate
0.2459	1.1765	20	0.4048	0.5613	0.6310	0.8812	0.9733	0.0102	0.9096	0.8391	0.0100	0.8349
0.2421	2.3529	40	0.1840	0.6645	0.7118	0.9292	0.9969	0.1720	0.9666	0.9574	0.1475	0.8886
0.1511	3.5294	60	0.1347	0.6751	0.7154	0.9392	0.9909	0.1590	0.9963	0.9639	0.1570	0.9045
0.1449	4.7059	80	0.1350	0.7359	0.7793	0.9471	0.9937	0.3623	0.9819	0.9642	0.3221	0.9213
0.1276	5.8824	100	0.1006	0.8194	0.9138	0.9551	0.9823	0.8117	0.9474	0.9707	0.5605	0.9271
0.0638	7.0588	120	0.0916	0.8139	0.8438	0.9646	0.9964	0.5438	0.9913	0.9779	0.5208	0.9431
0.0535	8.2353	140	0.0695	0.8572	0.8769	0.9735	0.9969	0.6367	0.9971	0.9804	0.6316	0.9597
0.0346	9.4118	160	0.0435	0.9224	0.9384	0.9848	0.9962	0.8230	0.9959	0.9888	0.8039	0.9745
0.0393	10.5882	180	0.0376	0.9352	0.9642	0.9867	0.9970	0.9082	0.9873	0.9882	0.8376	0.9798
0.0294	11.7647	200	0.0448	0.9298	0.9746	0.9851	0.9932	0.9487	0.9818	0.9916	0.8253	0.9725
0.0387	12.9412	220	0.0409	0.9270	0.9488	0.9855	0.9970	0.8575	0.9918	0.9830	0.8157	0.9823
0.0435	14.1176	240	0.0353	0.9482	0.9685	0.9886	0.9891	0.9185	0.9980	0.9881	0.8749	0.9816
0.022	15.2941	260	0.0246	0.9587	0.9696	0.9915	0.9970	0.9152	0.9967	0.9931	0.8979	0.9853
0.0203	16.4706	280	0.0191	0.9698	0.9826	0.9934	0.9953	0.9557	0.9967	0.9935	0.9272	0.9887
0.0212	17.6471	300	0.0256	0.9604	0.9724	0.9917	0.9953	0.9243	0.9975	0.9933	0.9028	0.9851
0.0123	18.8235	320	0.0223	0.9638	0.9763	0.9924	0.9954	0.9363	0.9972	0.9938	0.9112	0.9864
0.0137	20.0	340	0.0292	0.9543	0.9720	0.9906	0.9933	0.9256	0.9969	0.9919	0.8867	0.9844
0.0092	21.1765	360	0.0171	0.9719	0.9797	0.9941	0.9977	0.9439	0.9974	0.9942	0.9312	0.9902
0.0094	22.3529	380	0.0178	0.9730	0.9829	0.9941	0.9984	0.9550	0.9952	0.9938	0.9352	0.9901
0.016	23.5294	400	0.0163	0.9760	0.9881	0.9946	0.9954	0.9721	0.9969	0.9944	0.9430	0.9907
0.0083	24.7059	420	0.0151	0.9784	0.9882	0.9952	0.9973	0.9707	0.9965	0.9952	0.9483	0.9916
0.0094	25.8824	440	0.0259	0.9626	0.9731	0.9925	0.9971	0.9248	0.9972	0.9952	0.9067	0.9858
0.0144	27.0588	460	0.0171	0.9743	0.9860	0.9945	0.9980	0.9648	0.9951	0.9948	0.9376	0.9905
0.0075	28.2353	480	0.0168	0.9733	0.9824	0.9943	0.9972	0.9528	0.9972	0.9949	0.9351	0.9900
0.0076	29.4118	500	0.0171	0.9756	0.9842	0.9947	0.9979	0.9580	0.9966	0.9951	0.9409	0.9907
0.0075	30.5882	520	0.0170	0.9748	0.9835	0.9946	0.9974	0.9560	0.9971	0.9954	0.9388	0.9901
0.0084	31.7647	540	0.0154	0.9783	0.9899	0.9952	0.9976	0.9770	0.9953	0.9954	0.9480	0.9914
0.0055	32.9412	560	0.0156	0.9777	0.9888	0.9951	0.9971	0.9730	0.9962	0.9953	0.9465	0.9913
0.009	34.1176	580	0.0166	0.9752	0.9856	0.9947	0.9972	0.9630	0.9965	0.9953	0.9400	0.9904
0.0055	35.2941	600	0.0176	0.9745	0.9835	0.9946	0.9972	0.9560	0.9974	0.9954	0.9378	0.9902
0.0069	36.4706	620	0.0180	0.9748	0.9832	0.9946	0.9974	0.9547	0.9974	0.9955	0.9388	0.9902
0.0051	37.6471	640	0.0181	0.9752	0.9843	0.9947	0.9975	0.9585	0.9968	0.9955	0.9397	0.9903
0.0071	38.8235	660	0.0201	0.9729	0.9847	0.9943	0.9968	0.9610	0.9963	0.9953	0.9337	0.9896
0.0058	40.0	680	0.0208	0.9720	0.9826	0.9941	0.9971	0.9540	0.9968	0.9954	0.9315	0.9892
0.0061	41.1765	700	0.0222	0.9699	0.9802	0.9937	0.9973	0.9467	0.9967	0.9954	0.9260	0.9883
0.0062	42.3529	720	0.0205	0.9720	0.9819	0.9941	0.9975	0.9516	0.9966	0.9953	0.9315	0.9891
0.004	43.5294	740	0.0193	0.9741	0.9835	0.9945	0.9973	0.9561	0.9969	0.9954	0.9371	0.9898
0.0065	44.7059	760	0.0195	0.9738	0.9842	0.9944	0.9971	0.9588	0.9967	0.9953	0.9363	0.9898
0.0044	45.8824	780	0.0201	0.9731	0.9830	0.9943	0.9971	0.9550	0.9969	0.9954	0.9344	0.9895
0.0073	47.0588	800	0.0210	0.9723	0.9818	0.9941	0.9972	0.9512	0.9971	0.9954	0.9323	0.9891
0.0049	48.2353	820	0.0209	0.9723	0.9822	0.9941	0.9974	0.9527	0.9966	0.9954	0.9322	0.9892
0.0069	49.4118	840	0.0210	0.9721	0.9816	0.9941	0.9974	0.9506	0.9969	0.9954	0.9316	0.9891

Framework versions

• Transformers 4.41.2
• Pytorch 2.0.1+cu117
• Datasets 2.19.2
• Tokenizers 0.19.1