segformer-b1-finetuned-UBC

This model is a fine-tuned version of nvidia/segformer-b4-finetuned-ade-512-512 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6250
• Mean Iou: 0.3060
• Mean Accuracy: 0.5177
• Overall Accuracy: 0.6022
• Accuracy Road-trees-ocean: nan
• Accuracy Residential: 0.7533
• Accuracy Commercial: 0.4895
• Accuracy Industrial: 0.6393
• Accuracy Public: 0.4912
• Accuracy Other: 0.2152
• Iou Road-trees-ocean: 0.0
• Iou Residential: 0.6788
• Iou Commercial: 0.3656
• Iou Industrial: 0.2652
• Iou Public: 0.3602
• Iou Other: 0.1662

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: 5
• eval_batch_size: 5
• seed: 42
• optimizer: Use OptimizerNames.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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Road-trees-ocean	Accuracy Residential	Accuracy Commercial	Accuracy Industrial	Accuracy Public	Accuracy Other	Iou Road-trees-ocean	Iou Residential	Iou Commercial	Iou Industrial	Iou Public	Iou Other
0.6615	1.0	112	0.7051	0.1920	0.3218	0.5703	nan	0.7995	0.6999	0.0002	0.1094	0.0	0.0	0.6690	0.3774	0.0002	0.1052	0.0
0.4892	2.0	224	0.4453	0.2012	0.3142	0.5312	nan	0.7007	0.6561	0.0	0.2144	0.0	0.0	0.6245	0.3895	0.0	0.1934	0.0
0.3549	3.0	336	0.4100	0.2273	0.3564	0.5065	nan	0.7146	0.2978	0.3118	0.4577	0.0	0.0	0.6335	0.2514	0.1735	0.3055	0.0
0.3597	4.0	448	0.4001	0.2534	0.4062	0.5525	nan	0.7658	0.3541	0.4291	0.4818	0.0	0.0	0.6590	0.2975	0.2454	0.3184	0.0
0.276	5.0	560	0.3985	0.2382	0.3652	0.5385	nan	0.6930	0.5632	0.2008	0.3588	0.0103	0.0	0.6399	0.3665	0.1302	0.2827	0.0102
0.2568	6.0	672	0.3932	0.2715	0.4139	0.5838	nan	0.7988	0.4944	0.3519	0.3776	0.0467	0.0	0.6785	0.3799	0.2135	0.3129	0.0440
0.2238	7.0	784	0.4032	0.2991	0.4681	0.5846	nan	0.7494	0.5426	0.5155	0.3801	0.1527	0.0	0.6745	0.3865	0.3042	0.3037	0.1257
0.1856	8.0	896	0.4345	0.2946	0.4942	0.5812	nan	0.7566	0.3599	0.6136	0.5516	0.1892	0.0	0.6851	0.2924	0.2887	0.3528	0.1489
0.1733	9.0	1008	0.4259	0.2993	0.4755	0.5946	nan	0.7494	0.5293	0.5092	0.4660	0.1238	0.0	0.6848	0.3785	0.2872	0.3398	0.1054
0.177	10.0	1120	0.4989	0.2675	0.4962	0.5674	nan	0.7871	0.2486	0.7673	0.5283	0.1495	0.0	0.6960	0.2170	0.2421	0.3289	0.1211
0.128	11.0	1232	0.4857	0.2926	0.5225	0.5677	nan	0.6727	0.4868	0.7228	0.4968	0.2335	0.0	0.6400	0.3525	0.2497	0.3547	0.1586
0.1277	12.0	1344	0.4589	0.3046	0.5037	0.6014	nan	0.7587	0.4364	0.5857	0.5713	0.1663	0.0	0.6857	0.3491	0.2877	0.3670	0.1382
0.1465	13.0	1456	0.4877	0.2978	0.5149	0.5791	nan	0.7151	0.4300	0.6773	0.5372	0.2151	0.0	0.6608	0.3373	0.2732	0.3567	0.1589
0.1692	14.0	1568	0.4828	0.3006	0.4966	0.5947	nan	0.7469	0.4576	0.5661	0.5380	0.1745	0.0	0.6808	0.3505	0.2721	0.3595	0.1407
0.1402	15.0	1680	0.5083	0.3008	0.5034	0.5905	nan	0.7414	0.4931	0.6033	0.4570	0.2224	0.0	0.6659	0.3629	0.2614	0.3504	0.1642
0.1815	16.0	1792	0.5092	0.2995	0.5242	0.5847	nan	0.7432	0.4522	0.6954	0.4245	0.3056	0.0	0.6767	0.3478	0.2400	0.3362	0.1965
0.1327	17.0	1904	0.5164	0.3064	0.5051	0.5940	nan	0.7289	0.5073	0.5805	0.5011	0.2077	0.0	0.6639	0.3757	0.2818	0.3643	0.1528
0.1045	18.0	2016	0.5242	0.3021	0.5083	0.6035	nan	0.7820	0.4367	0.6433	0.5077	0.1717	0.0	0.6898	0.3522	0.2736	0.3597	0.1375
0.0935	19.0	2128	0.5375	0.2995	0.5132	0.5826	nan	0.7114	0.4796	0.6702	0.5107	0.1942	0.0	0.6584	0.3477	0.2836	0.3558	0.1516
0.1029	20.0	2240	0.5384	0.2934	0.5002	0.5858	nan	0.7347	0.3958	0.6215	0.6049	0.1442	0.0	0.6786	0.3180	0.2778	0.3658	0.1203
0.1076	21.0	2352	0.5209	0.3030	0.4923	0.5998	nan	0.7554	0.4506	0.4999	0.5542	0.2013	0.0	0.6809	0.3515	0.2704	0.3588	0.1565
0.1118	22.0	2464	0.5442	0.2876	0.4962	0.5899	nan	0.7744	0.4817	0.6387	0.3802	0.2060	0.0	0.6798	0.3616	0.2201	0.3017	0.1627
0.0962	23.0	2576	0.5447	0.3073	0.5146	0.6054	nan	0.7627	0.4891	0.6101	0.4858	0.2255	0.0	0.6828	0.3689	0.2692	0.3511	0.1717
0.098	24.0	2688	0.5635	0.2999	0.5139	0.5964	nan	0.7423	0.4949	0.6486	0.4872	0.1964	0.0	0.6779	0.3582	0.2564	0.3482	0.1588
0.1051	25.0	2800	0.5509	0.3002	0.5153	0.5976	nan	0.7759	0.4163	0.6716	0.4958	0.2169	0.0	0.6932	0.3270	0.2636	0.3516	0.1657
0.1073	26.0	2912	0.5476	0.3103	0.4987	0.6083	nan	0.7506	0.5518	0.4992	0.4737	0.2181	0.0	0.6793	0.3950	0.2688	0.3563	0.1627
0.1398	27.0	3024	0.5794	0.3043	0.5146	0.5989	nan	0.7554	0.4470	0.6395	0.5288	0.2025	0.0	0.6811	0.3501	0.2817	0.3538	0.1591
0.1013	28.0	3136	0.5661	0.3042	0.5002	0.6027	nan	0.7400	0.5420	0.5505	0.4905	0.1781	0.0	0.6683	0.3882	0.2677	0.3551	0.1456
0.0693	29.0	3248	0.5628	0.3096	0.5212	0.6097	nan	0.7531	0.5093	0.6121	0.5100	0.2214	0.0	0.6836	0.3813	0.2634	0.3599	0.1693
0.0954	30.0	3360	0.5839	0.2934	0.5060	0.5921	nan	0.7682	0.4295	0.6672	0.4823	0.1827	0.0	0.6774	0.3386	0.2436	0.3560	0.1446
0.0792	31.0	3472	0.5779	0.3021	0.5069	0.6012	nan	0.7463	0.5260	0.5756	0.4653	0.2212	0.0	0.6793	0.3729	0.2372	0.3542	0.1692
0.0842	32.0	3584	0.5680	0.3155	0.5169	0.6178	nan	0.7789	0.5031	0.5572	0.4906	0.2548	0.0	0.6903	0.3851	0.2612	0.3726	0.1839
0.0778	33.0	3696	0.5928	0.3053	0.5157	0.6025	nan	0.7550	0.4850	0.6293	0.4979	0.2115	0.0	0.6814	0.3648	0.2582	0.3644	0.1630
0.0681	34.0	3808	0.5967	0.3054	0.5124	0.6045	nan	0.7600	0.4790	0.6243	0.5157	0.1831	0.0	0.6832	0.3673	0.2687	0.3655	0.1480
0.1204	35.0	3920	0.6043	0.3051	0.5284	0.6009	nan	0.7611	0.4233	0.7087	0.5413	0.2079	0.0	0.6865	0.3399	0.2723	0.3688	0.1631
0.063	36.0	4032	0.5979	0.3065	0.5126	0.6024	nan	0.7561	0.4720	0.6008	0.5127	0.2210	0.0	0.6801	0.3614	0.2639	0.3652	0.1682
0.0861	37.0	4144	0.6206	0.3025	0.5210	0.5957	nan	0.7405	0.4709	0.6698	0.5083	0.2151	0.0	0.6742	0.3580	0.2621	0.3568	0.1638
0.0716	38.0	4256	0.6022	0.3003	0.5082	0.5999	nan	0.7575	0.4981	0.6141	0.4636	0.2075	0.0	0.6791	0.3709	0.2434	0.3476	0.1610
0.0698	39.0	4368	0.6149	0.3055	0.5139	0.6070	nan	0.7747	0.4750	0.6323	0.4872	0.2002	0.0	0.6845	0.3681	0.2625	0.3598	0.1578
0.0966	40.0	4480	0.6190	0.3050	0.5135	0.6049	nan	0.7670	0.4782	0.6261	0.4913	0.2051	0.0	0.6819	0.3670	0.2601	0.3618	0.1594
0.0792	41.0	4592	0.6239	0.3038	0.5170	0.5996	nan	0.7436	0.4801	0.6419	0.5223	0.1970	0.0	0.6773	0.3643	0.2615	0.3643	0.1555
0.0847	42.0	4704	0.6188	0.3061	0.5182	0.6043	nan	0.7546	0.4909	0.6345	0.5010	0.2099	0.0	0.6802	0.3715	0.2567	0.3647	0.1633
0.0699	43.0	4816	0.6188	0.3070	0.5141	0.6078	nan	0.7694	0.4687	0.6024	0.5124	0.2177	0.0	0.6868	0.3643	0.2598	0.3639	0.1671
0.073	44.0	4928	0.6249	0.3042	0.5186	0.5993	nan	0.7432	0.4952	0.6594	0.4987	0.1966	0.0	0.6770	0.3678	0.2628	0.3620	0.1558
0.0707	45.0	5040	0.6273	0.3066	0.5247	0.5990	nan	0.7430	0.4767	0.6635	0.5071	0.2332	0.0	0.6734	0.3602	0.2637	0.3665	0.1759
0.0716	46.0	5152	0.6314	0.3074	0.5220	0.6055	nan	0.7554	0.4911	0.6483	0.5007	0.2147	0.0	0.6818	0.3675	0.2633	0.3644	0.1672
0.1249	47.0	5264	0.6242	0.3081	0.5194	0.6054	nan	0.7627	0.4854	0.6419	0.4876	0.2194	0.0	0.6810	0.3687	0.2679	0.3632	0.1674
0.0772	48.0	5376	0.6427	0.3065	0.5249	0.6022	nan	0.7516	0.4817	0.6720	0.4979	0.2215	0.0	0.6789	0.3633	0.2653	0.3622	0.1696
0.0914	49.0	5488	0.6323	0.3067	0.5234	0.6040	nan	0.7571	0.4809	0.6677	0.4985	0.2127	0.0	0.6816	0.3634	0.2673	0.3631	0.1649
0.0679	50.0	5600	0.6250	0.3060	0.5177	0.6022	nan	0.7533	0.4895	0.6393	0.4912	0.2152	0.0	0.6788	0.3656	0.2652	0.3602	0.1662

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1