segformer-b0-scene-parse-150-lr-4-e-30-new-9img

This model is a fine-tuned version of DiTo97/binarization-segformer-b3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1623
• Mean Iou: 0.5077
• Mean Accuracy: 0.5342
• Overall Accuracy: 0.9575
• Per Category Iou: [0.05804449685867197, 0.9574223586207944]
• Per Category Accuracy: [0.0820083385013113, 0.9863919593485183]

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: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
No log	1.0	132	0.2043	0.4839	0.4998	0.9662	[0.001549543069893383, 0.9662390544491327]	[0.001641922637237834, 0.9980349550926039]
No log	2.0	264	0.1582	0.4840	0.5000	0.9680	[6.710408589322995e-05, 0.9680253773280134]	[6.724597831690788e-05, 0.999930303583307]
No log	3.0	396	0.1382	0.4841	0.5001	0.9681	[0.0001139354228965373, 0.9680917294976413]	[0.00011394457437031613, 0.9999973525212741]
0.3872	4.0	528	0.1699	0.4951	0.5117	0.9660	[0.024248402101243276, 0.9659487595053663]	[0.02649678340070384, 0.9969437628725957]
0.3872	5.0	660	0.1275	0.4860	0.5019	0.9680	[0.003997176715331179, 0.9680342203165365]	[0.004019815148277382, 0.9998133219651828]
0.3872	6.0	792	0.1469	0.4942	0.5106	0.9665	[0.021996863056051138, 0.9664511446344629]	[0.023629489603024575, 0.9975536065186591]
0.3872	7.0	924	0.1472	0.5057	0.5257	0.9633	[0.048193367074049226, 0.9632320238098082]	[0.058235017222442224, 0.9931321939077451]
0.1503	8.0	1056	0.1385	0.4982	0.5147	0.9668	[0.02966123083403392, 0.9667378174750562]	[0.0318316310138452, 0.997588146880642]
0.1503	9.0	1188	0.1401	0.4950	0.5115	0.9663	[0.023698989853648766, 0.9662662110495017]	[0.025641265121005404, 0.9972986481604209]
0.1503	10.0	1320	0.1440	0.4968	0.5141	0.9649	[0.02877413033473597, 0.9648685554910891]	[0.03258441238222614, 0.9956352926892399]
0.1503	11.0	1452	0.1532	0.5117	0.5385	0.9588	[0.06479279776445032, 0.9586699704883279]	[0.08948011386985662, 0.9874412228938199]
0.1334	12.0	1584	0.1425	0.5083	0.5310	0.9612	[0.05548374916494815, 0.9611461513466955]	[0.0713647944888185, 0.9905656172060997]
0.1334	13.0	1716	0.1457	0.5052	0.5266	0.9615	[0.04888450573004655, 0.9614673387061856]	[0.06194288574908284, 0.9911952245886988]
0.1334	14.0	1848	0.1510	0.5087	0.5322	0.9605	[0.05697498023630964, 0.960444678467155]	[0.07471588574161106, 0.9897365869492352]
0.1334	15.0	1980	0.1445	0.5063	0.5262	0.9635	[0.049058901751704166, 0.9634450432339506]	[0.0589896665346653, 0.993327861056376]
0.1218	16.0	2112	0.1488	0.5037	0.5242	0.9621	[0.04539026025348957, 0.9619926039074217]	[0.05653332038225603, 0.9919082583358022]
0.1218	17.0	2244	0.1554	0.5125	0.5458	0.9542	[0.07088184464528238, 0.9540476640217492]	[0.1094801885876103, 0.98205126405411]
0.1218	18.0	2376	0.1484	0.5048	0.5257	0.9621	[0.04769031825474748, 0.9620025903307378]	[0.059520162585832016, 0.9918238468629347]
0.1136	19.0	2508	0.1478	0.5065	0.5278	0.9621	[0.05100673744263626, 0.9620083632427995]	[0.063863132018799, 0.9916920886193632]
0.1136	20.0	2640	0.1482	0.5102	0.5366	0.9585	[0.06195357859711988, 0.9583808144125242]	[0.08590673729984982, 0.9872562687986381]
0.1136	21.0	2772	0.1479	0.5043	0.5255	0.9615	[0.04721944072757992, 0.9614743147367695]	[0.059721900520782745, 0.9912728018722969]
0.1136	22.0	2904	0.1528	0.5084	0.5334	0.9590	[0.057761084324792794, 0.958942151209778]	[0.07867966257462436, 0.9880629494095138]
0.1053	23.0	3036	0.1642	0.5118	0.5549	0.9476	[0.07620148345702486, 0.9474144999535162]	[0.13534934285735634, 0.9744155659927787]
0.1053	24.0	3168	0.1583	0.5085	0.5367	0.9564	[0.060741142425379466, 0.9563097838863447]	[0.08829583747394218, 0.9850475327099075]
0.1053	25.0	3300	0.1627	0.5101	0.5380	0.9573	[0.0631439155292623, 0.9571557211026566]	[0.09023663112592183, 0.9858576612000541]
0.1053	26.0	3432	0.1646	0.5101	0.5441	0.9527	[0.06773853824581035, 0.9525613225447881]	[0.10765147156615884, 0.9805787117590169]
0.1024	27.0	3564	0.1616	0.5089	0.5403	0.9540	[0.06385837245316543, 0.9538794921181668]	[0.09828560114168727, 0.9822301227912947]
0.1024	28.0	3696	0.1614	0.5077	0.5352	0.9566	[0.058942135207107095, 0.9565120386869382]	[0.0851296726615211, 0.9853556869197558]
0.1024	29.0	3828	0.1628	0.5076	0.5353	0.9564	[0.05886607703633537, 0.9562951940049768]	[0.08543601545163146, 0.9851226472225978]
0.1024	30.0	3960	0.1623	0.5077	0.5342	0.9575	[0.05804449685867197, 0.9574223586207944]	[0.0820083385013113, 0.9863919593485183]

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2