segformer-b0-finetuned-busigt2

This model is a fine-tuned version of nvidia/mit-b1 on the kasumi222/busigt5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2904
• Mean Iou: 0.4458
• Mean Accuracy: 0.6980
• Overall Accuracy: 0.6969
• Per Category Iou: [0.0, 0.6551336334577343, 0.6821319425157643]
• Per Category Accuracy: [nan, 0.6913100552356098, 0.70464740289276]

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: 0.00013
• train_batch_size: 20
• eval_batch_size: 20
• 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	Per Category Iou	Per Category Accuracy
0.1095	0.77	20	0.2086	0.4674	0.7410	0.7419	[0.0, 0.6978460673452154, 0.704309291034096]	[nan, 0.7461995349612959, 0.7357650020760118]
0.1156	1.54	40	0.1980	0.4186	0.6721	0.6783	[0.0, 0.6446507442278364, 0.6112330250576428]	[nan, 0.7089917293749448, 0.635300900559587]
0.1039	2.31	60	0.1987	0.3706	0.5810	0.5757	[0.0, 0.5345322994102119, 0.5773860979625277]	[nan, 0.5495831330265778, 0.6123860258526792]
0.0672	3.08	80	0.1960	0.4099	0.6407	0.6439	[0.0, 0.6194380206711395, 0.6103561290824698]	[nan, 0.6596136450596995, 0.6218662960315686]
0.0992	3.85	100	0.1969	0.4201	0.6684	0.6695	[0.0, 0.6251984513525223, 0.6351366565306488]	[nan, 0.675036447653713, 0.661700391303438]
0.085	4.62	120	0.2075	0.4383	0.6997	0.6964	[0.0, 0.6407576836532538, 0.6742246105299582]	[nan, 0.6804532655724195, 0.718889834811138]
0.0561	5.38	140	0.2037	0.4401	0.7033	0.7071	[0.0, 0.6545188689920507, 0.665783897448558]	[nan, 0.7263735810923504, 0.6801427547189345]
0.0841	6.15	160	0.2119	0.3651	0.5891	0.5934	[0.0, 0.5494216923933923, 0.5458843877102458]	[nan, 0.6146571565924632, 0.5634664881039569]
0.1034	6.92	180	0.2371	0.3684	0.6193	0.6367	[0.0, 0.6047004430113216, 0.5003660220404046]	[nan, 0.7229919452156935, 0.5156554415186935]
0.0691	7.69	200	0.2266	0.4285	0.6991	0.7117	[0.0, 0.6730686627556878, 0.6124621276402561]	[nan, 0.7742042834577688, 0.6240342690621383]
0.0601	8.46	220	0.2106	0.4198	0.6674	0.6704	[0.0, 0.6308213023617786, 0.6287108585057931]	[nan, 0.6851880267250091, 0.6497046776895365]
0.0647	9.23	240	0.2234	0.4229	0.6746	0.6777	[0.0, 0.6338885508159525, 0.6349404984513296]	[nan, 0.6928998204597407, 0.6563077167064432]
0.0626	10.0	260	0.2322	0.3991	0.6540	0.6655	[0.0, 0.6267222060572648, 0.570544858752452]	[nan, 0.7227113522422911, 0.5852409330048426]
0.0604	10.77	280	0.2021	0.4660	0.7283	0.7288	[0.0, 0.6990308020264264, 0.6989818924111941]	[nan, 0.7310753774760368, 0.7255727204344536]
0.0573	11.54	300	0.2227	0.4513	0.7014	0.6951	[0.0, 0.6488805486358904, 0.7049138389320693]	[nan, 0.6638350976679388, 0.7389417956785915]
0.0474	12.31	320	0.2108	0.4781	0.7468	0.7371	[0.0, 0.6761855871787447, 0.7580093480444655]	[nan, 0.6890590324447889, 0.8044529075728725]
0.0805	13.08	340	0.2257	0.4325	0.6902	0.6940	[0.0, 0.6550347525850334, 0.6423545682885212]	[nan, 0.7128733309133007, 0.6675247882412931]
0.0545	13.85	360	0.2155	0.4609	0.7230	0.7167	[0.0, 0.6629649481906197, 0.7196967289093881]	[nan, 0.6853650161390015, 0.7606061073292577]
0.0628	14.62	380	0.2397	0.4150	0.6561	0.6611	[0.0, 0.6377593821077956, 0.6070948266377257]	[nan, 0.6861969841160831, 0.6259296622984148]
0.0576	15.38	400	0.2177	0.4661	0.7274	0.7272	[0.0, 0.6936915190759695, 0.7046022162863222]	[nan, 0.7263017649886684, 0.7284576609239519]
0.0808	16.15	420	0.2263	0.4248	0.6707	0.6740	[0.0, 0.6438773235874202, 0.6304024210524071]	[nan, 0.6904172594111472, 0.6510802419847774]
0.0458	16.92	440	0.2342	0.4006	0.6449	0.6525	[0.0, 0.6208902028936363, 0.5809796433249929]	[nan, 0.6898132977523129, 0.6000533044931062]
0.0477	17.69	460	0.2683	0.3789	0.6170	0.6232	[0.0, 0.5741692028709614, 0.5625631837395161]	[nan, 0.6539633266945951, 0.5800762342358019]
0.0501	18.46	480	0.2364	0.4280	0.6700	0.6675	[0.0, 0.6223049989658083, 0.6617065588280534]	[nan, 0.6552936905824757, 0.6846169180090992]
0.039	19.23	500	0.2378	0.4500	0.7052	0.6986	[0.0, 0.6391919313721981, 0.7106968345576296]	[nan, 0.665670921345669, 0.7446979100013106]
0.041	20.0	520	0.2477	0.4142	0.6612	0.6659	[0.0, 0.6273087938535062, 0.6153514032911991]	[nan, 0.6890233206118104, 0.6333526433632052]
0.0331	20.77	540	0.2488	0.4353	0.6814	0.6778	[0.0, 0.6267198588955959, 0.6791644212315564]	[nan, 0.6603973431966015, 0.7023153313193633]
0.0316	21.54	560	0.2468	0.4500	0.7025	0.6974	[0.0, 0.6405571933079939, 0.7093320446678179]	[nan, 0.6719456081313097, 0.7331179494069875]
0.0333	22.31	580	0.2477	0.4384	0.6899	0.6906	[0.0, 0.6520329743081146, 0.6630535380613215]	[nan, 0.6937796658392771, 0.6860558089232162]
0.0269	23.08	600	0.2603	0.4477	0.7018	0.6996	[0.0, 0.6514078130357787, 0.6916101875532822]	[nan, 0.6888588892050193, 0.7147725032516842]
0.033	23.85	620	0.2424	0.4499	0.7061	0.6986	[0.0, 0.6447352671115818, 0.7048670621273163]	[nan, 0.6616131152687708, 0.750523958937919]
0.0555	24.62	640	0.2471	0.4342	0.6830	0.6823	[0.0, 0.636756610371055, 0.6659104633164847]	[nan, 0.6791280033749645, 0.6868014110272018]
0.0583	25.38	660	0.2517	0.4434	0.6922	0.6879	[0.0, 0.6386719513699022, 0.6913843141331489]	[nan, 0.6666374954624388, 0.7178391636040445]
0.154	26.15	680	0.2535	0.4235	0.6597	0.6487	[0.0, 0.5750726006840868, 0.695285501846172]	[nan, 0.5943477194462704, 0.7250215035171054]
0.0292	26.92	700	0.2768	0.3679	0.6035	0.6135	[0.0, 0.5756677002657924, 0.5279750019379379]	[nan, 0.6631412677700708, 0.5438385402498483]
0.0288	27.69	720	0.2455	0.4676	0.7235	0.7188	[0.0, 0.6761224569996822, 0.7268002447671437]	[nan, 0.6954373227898398, 0.7515024928661187]
0.0321	28.46	740	0.2618	0.4324	0.6745	0.6691	[0.0, 0.6201514037000198, 0.6770266576179022]	[nan, 0.6425218048210974, 0.7064552401951121]
0.0309	29.23	760	0.2742	0.3944	0.6348	0.6407	[0.0, 0.6008533572398147, 0.5822751024176394]	[nan, 0.6701804232440864, 0.599451426280657]
0.0244	30.0	780	0.2667	0.4386	0.6819	0.6750	[0.0, 0.6224630782821559, 0.693390305711243]	[nan, 0.6412495217165226, 0.7224713681082742]
0.0642	30.77	800	0.2501	0.4581	0.7121	0.7096	[0.0, 0.6722145834845955, 0.7021141065136746]	[nan, 0.6976031865943273, 0.7265325317101161]
0.0481	31.54	820	0.2685	0.4137	0.6689	0.6766	[0.0, 0.6379976664903103, 0.6031984018650592]	[nan, 0.7145859291453688, 0.6231961550279683]
0.0311	32.31	840	0.2570	0.4284	0.6804	0.6832	[0.0, 0.6426329055663264, 0.6425854743219936]	[nan, 0.6969752862342657, 0.6639063603053335]
0.0389	33.08	860	0.2795	0.3918	0.6456	0.6590	[0.0, 0.6244554318979076, 0.5508200429573112]	[nan, 0.7254125011037311, 0.5658618862962298]
0.0282	33.85	880	0.2568	0.4242	0.6759	0.6775	[0.0, 0.6282787291971401, 0.6442735430594793]	[nan, 0.6857107537747603, 0.6660974613184492]
0.0245	34.62	900	0.2635	0.4503	0.7043	0.7037	[0.0, 0.6658605581388065, 0.6850412042515538]	[nan, 0.7008356961354695, 0.7076892832638209]
0.0315	35.38	920	0.2769	0.4443	0.7038	0.7055	[0.0, 0.6610872730365329, 0.6718978137221756]	[nan, 0.7138198907060935, 0.6938235070611933]
0.0283	36.15	940	0.2697	0.4392	0.6920	0.6907	[0.0, 0.6405508279799802, 0.6769668218170816]	[nan, 0.6841213809883544, 0.6998318265269149]
0.0257	36.92	960	0.2712	0.4562	0.7099	0.7082	[0.0, 0.6720494469697227, 0.6964887349332429]	[nan, 0.6999154296702542, 0.7197879714666775]
0.0188	37.69	980	0.2857	0.4300	0.6763	0.6771	[0.0, 0.6397832221652129, 0.6501046733477022]	[nan, 0.6811686795451647, 0.6713607293464362]
0.0259	38.46	1000	0.2812	0.4368	0.6851	0.6838	[0.0, 0.6396217765000503, 0.6707000380577134]	[nan, 0.6772780519391329, 0.6929027930893589]
0.0169	39.23	1020	0.2795	0.4542	0.7084	0.7054	[0.0, 0.6598929743362643, 0.7028156867427239]	[nan, 0.6906225043413423, 0.7260947520404938]
0.0296	40.0	1040	0.2834	0.4470	0.7015	0.7013	[0.0, 0.6608002641121026, 0.6801095152287282]	[nan, 0.7006602764723773, 0.7022773353480376]
0.0183	40.77	1060	0.2874	0.4386	0.6909	0.6903	[0.0, 0.6432231900832152, 0.6726091072738183]	[nan, 0.6874296310104291, 0.694422081276136]
0.0199	41.54	1080	0.2741	0.4594	0.7175	0.7154	[0.0, 0.6721657359810768, 0.7061664449453671]	[nan, 0.7051238631569653, 0.7298866398455491]
0.0162	42.31	1100	0.2883	0.4414	0.6921	0.6913	[0.0, 0.6492915338226911, 0.6750215527697642]	[nan, 0.6870752597447193, 0.6971930338516571]
0.0179	43.08	1120	0.2927	0.4425	0.6936	0.6927	[0.0, 0.651082790586508, 0.6764744769464034]	[nan, 0.6884633119781804, 0.6987260886947118]
0.0228	43.85	1140	0.2954	0.4273	0.6807	0.6841	[0.0, 0.6418083531582984, 0.6399672125377378]	[nan, 0.7006630235364526, 0.6608033559804007]
0.0164	44.62	1160	0.2954	0.4264	0.6740	0.6756	[0.0, 0.6356634502412776, 0.6436554266840772]	[nan, 0.6834636553611899, 0.6644801545389767]
0.0158	45.38	1180	0.2906	0.4433	0.6956	0.6951	[0.0, 0.6536928350497138, 0.6760836624911459]	[nan, 0.6927067410990219, 0.6985223421818058]
0.0198	46.15	1200	0.2881	0.4441	0.6969	0.6961	[0.0, 0.6527988151987781, 0.6794425179962712]	[nan, 0.6919179412716945, 0.7019810769049473]
0.018	46.92	1220	0.2961	0.4350	0.6844	0.6839	[0.0, 0.6395287774950378, 0.6655290939553297]	[nan, 0.6815206961845243, 0.6872821426644097]
0.0179	47.69	1240	0.2898	0.4459	0.6987	0.6982	[0.0, 0.6581945977423002, 0.6796217960953337]	[nan, 0.6955130632707722, 0.701934270273604]
0.0213	48.46	1260	0.2902	0.4469	0.7004	0.6998	[0.0, 0.6595482974648909, 0.6811920247361126]	[nan, 0.6971510983350829, 0.7036303223269834]
0.0227	49.23	1280	0.2888	0.4452	0.6967	0.6953	[0.0, 0.6532891096762087, 0.6823149709479772]	[nan, 0.6885578894699147, 0.7047801134592744]
0.0266	50.0	1300	0.2904	0.4458	0.6980	0.6969	[0.0, 0.6551336334577343, 0.6821319425157643]	[nan, 0.6913100552356098, 0.70464740289276]

Framework versions

• Transformers 4.21.3
• Pytorch 1.12.1+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1