segmentation_test

This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1208
• Mean Iou: 0.8483
• Mean Accuracy: 0.8963
• Overall Accuracy: 0.9648
• Per Category Iou: [0.896665866220625, 0.9313625624392157, 0.6184457649558176, 0.946788356429534]
• Per Category Accuracy: [0.9488350594956452, 0.9690753342283529, 0.6948783145659281, 0.9725778146880949]

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
1.0467	0.12	20	0.9970	0.4693	0.5892	0.8499	[0.016665768511477424, 0.7530905304669141, 0.32197766779593034, 0.785573865597306]	[0.018291186933739902, 0.9736110066215201, 0.5554138836791528, 0.8094802372221535]
0.7279	0.25	40	0.5469	0.4852	0.5502	0.8940	[0.01804951884861912, 0.8184740635118999, 0.24534751963740423, 0.8590701787560616]	[0.018703012459912727, 0.9597806760541293, 0.3156729455312819, 0.9066840327022103]
0.7806	0.38	60	0.4260	0.5242	0.6073	0.9049	[0.01807194500339603, 0.8366012049171444, 0.3657984221980414, 0.8765049443865006]	[0.018709584143841017, 0.9630852108788323, 0.5316352476333795, 0.9157347069923247]
0.5022	0.5	80	0.3776	0.5673	0.6442	0.9155	[0.13278193831665586, 0.8571556379842626, 0.39241878205971326, 0.8868105485330777]	[0.13725838108756988, 0.9454001415642639, 0.5529329784947034, 0.9412691031815928]
0.4512	0.62	100	0.3027	0.6147	0.6726	0.9272	[0.23169125963788326, 0.8698512434242224, 0.4538044468559813, 0.9035720315308671]	[0.2351282792702802, 0.939292533650652, 0.5538645534494219, 0.9621157857635089]
0.5998	0.75	120	0.2814	0.6595	0.7168	0.9328	[0.3932326160264655, 0.873639653244202, 0.45899785777309354, 0.9119874363843206]	[0.40385626412912046, 0.9319378939772671, 0.5617421687163787, 0.9697880271073833]
0.6628	0.88	140	0.2398	0.6224	0.6878	0.9322	[0.23667963537424397, 0.8731766506602201, 0.4613980137799232, 0.9182017221798935]	[0.23933415698438568, 0.9457139193910193, 0.6013402818446667, 0.9646299239546201]
0.2891	1.0	160	0.2509	0.6993	0.8005	0.9365	[0.5407886628522588, 0.8871068137852166, 0.4540803897685749, 0.9151364246964913]	[0.6209277465257698, 0.962521364442151, 0.6770795495822739, 0.9413993019486105]
0.3733	1.12	180	0.2347	0.7473	0.8210	0.9448	[0.6897757129718597, 0.8989510444856402, 0.478662489496792, 0.9219419098227718]	[0.7343878695476929, 0.9509685390377233, 0.6376642920153895, 0.9610472258827709]
0.6024	1.25	200	0.2099	0.7563	0.8150	0.9486	[0.7054491243993367, 0.9058582674210225, 0.4870865705869561, 0.9266736934314719]	[0.7287186968788882, 0.9590880866678464, 0.6088175916934154, 0.9633364922034189]
0.3624	1.38	220	0.2075	0.7570	0.8244	0.9477	[0.6637631174949002, 0.8996125590304884, 0.5362553031346449, 0.9285364229598405]	[0.6913981038501306, 0.971192970317068, 0.6830841308553933, 0.9520221263504635]
0.6915	1.5	240	0.1703	0.7974	0.8482	0.9540	[0.8079702418058335, 0.9109995197988137, 0.5379447825240211, 0.9325512132433632]	[0.8631205860189615, 0.9445488751680811, 0.6074684141727886, 0.977596233163672]
0.3712	1.62	260	0.1817	0.7999	0.8580	0.9529	[0.8010650098967221, 0.9085553502562573, 0.560168977421881, 0.9298580626286841]	[0.8874336259923242, 0.9404067164614021, 0.6274738009987867, 0.9766288563247306]
0.2538	1.75	280	0.1751	0.8150	0.8836	0.9566	[0.8004213101386246, 0.9180591806306987, 0.6065608191143141, 0.9348731824073968]	[0.8772891365683543, 0.9692263290457869, 0.7281184904037895, 0.9597196634547871]
0.2284	1.88	300	0.1840	0.8126	0.9042	0.9567	[0.8009792424281661, 0.9194293484770127, 0.5939762508191415, 0.9359433726281363]	[0.9485721921385136, 0.9659392778315237, 0.7435920245322974, 0.9586582645062351]
0.256	2.0	320	0.1616	0.8176	0.8884	0.9579	[0.809949528422126, 0.9207851938667173, 0.6025298091308882, 0.937268997247811]	[0.8700449503180695, 0.9619340740732891, 0.7553170196520308, 0.9663523606431447]
0.4148	2.12	340	0.1642	0.8206	0.8719	0.9582	[0.7932048029906043, 0.9191135080588984, 0.6322722187829223, 0.937770246889229]	[0.8329324606137076, 0.9742867746553344, 0.7192549377179132, 0.9612368510870202]
0.2985	2.25	360	0.1486	0.8375	0.8977	0.9613	[0.8586827106799612, 0.9252188318786676, 0.6245207917181071, 0.9415713272802116]	[0.9194727757040464, 0.969478781784409, 0.7362456788018513, 0.9657594912576178]
0.1704	2.38	380	0.1467	0.8290	0.9022	0.9601	[0.8357054588123389, 0.9237357509430961, 0.6162257083130753, 0.9403938047508744]	[0.9392075425407006, 0.9538586328240481, 0.742260144159371, 0.9735591413949309]
0.1934	2.5	400	0.1474	0.8291	0.8797	0.9599	[0.8571529421535651, 0.9213274361346364, 0.5977020515839274, 0.9403603679518928]	[0.9043119008814818, 0.9737894189189619, 0.6774254925362808, 0.9630940434994081]
0.1912	2.62	420	0.1433	0.8310	0.9069	0.9606	[0.8355815564504465, 0.9245037970991156, 0.6227678339008111, 0.9410534282338662]	[0.9617834673956854, 0.9520821920420963, 0.7389588599697052, 0.9748254174032972]
0.1921	2.75	440	0.1339	0.8286	0.8766	0.9608	[0.8511074855171785, 0.9240321564072764, 0.5977901602255353, 0.941451956991874]	[0.9004806091512889, 0.9682032729318744, 0.6685767659770047, 0.9691355452863861]
0.139	2.88	460	0.1375	0.8324	0.8971	0.9615	[0.8342633614251669, 0.9255032673803006, 0.6270579504122933, 0.9426635896658996]	[0.9409074181168182, 0.9556867297277537, 0.7161216829630508, 0.9756464136284629]
0.1557	3.0	480	0.1368	0.8347	0.8932	0.9603	[0.8546089056727355, 0.9225367596519559, 0.6212372144507865, 0.9405009897594654]	[0.9146294446488968, 0.9608790971251646, 0.7266902402079611, 0.9707082533928404]
0.2095	3.12	500	0.1299	0.8431	0.9034	0.9629	[0.8774828339164978, 0.9283559261158525, 0.6227224847390085, 0.9439444287208694]	[0.9498317648914358, 0.9668135908051043, 0.727485909002177, 0.9695951469339585]
0.1639	3.25	520	0.1300	0.8384	0.8794	0.9628	[0.8790845930676439, 0.9277390439822828, 0.6029040896735798, 0.9440283579425929]	[0.9080161400557278, 0.9708537617842183, 0.6681270401367957, 0.9706067913536859]
0.1139	3.38	540	0.1235	0.8355	0.8815	0.9632	[0.8830324246537594, 0.928376002336871, 0.5856632431257472, 0.9450535716038398]	[0.9476412035820059, 0.9631465359669655, 0.639146904675419, 0.9760456588476107]
0.2574	3.5	560	0.1264	0.8453	0.9043	0.9638	[0.878600355282667, 0.9295598939555374, 0.6275502561490758, 0.945452356561073]	[0.9566728878607854, 0.9609682370480884, 0.7246467057582204, 0.9750175535837675]
0.2927	3.62	580	0.1309	0.8446	0.8981	0.9633	[0.8866855291248271, 0.9282894217693001, 0.6190576519555571, 0.944518781436582]	[0.9452994935422253, 0.9598107425660218, 0.7115206416747593, 0.9759016962109369]
0.4659	3.75	600	0.1202	0.8484	0.9041	0.9641	[0.8833081778126276, 0.9297489957164383, 0.6347412766855173, 0.9456458562459887]	[0.9525152024954875, 0.9567847534440064, 0.7285978684971991, 0.9785108795019711]
0.1478	3.88	620	0.1207	0.8503	0.9115	0.9644	[0.8839341322345905, 0.930666904641919, 0.640631018371209, 0.9461269754236669]	[0.9544604209382612, 0.9586230491204685, 0.7562115292902486, 0.9766407602120007]
0.2361	4.0	640	0.1206	0.8509	0.9104	0.9639	[0.8843630597662286, 0.9290525728561428, 0.6450059707401802, 0.9453487498415234]	[0.9602916951439637, 0.9520253703081671, 0.7487638716947004, 0.9805396552894644]
0.22	4.12	660	0.1193	0.8448	0.8875	0.9643	[0.8981674375904806, 0.9304938750380576, 0.6046215297860035, 0.9459224283816728]	[0.9472622364754745, 0.9670648514793784, 0.6612946667951598, 0.9744785306883144]
0.209	4.25	680	0.1181	0.8503	0.9019	0.9645	[0.8927503260337429, 0.9310490279930463, 0.6313541481285349, 0.9459033573690366]	[0.9507189422217549, 0.9618490401497869, 0.7188299220887047, 0.9762163122315232]
0.2477	4.38	700	0.1210	0.8483	0.8967	0.9646	[0.8965441720351905, 0.9309945711672964, 0.619250454020142, 0.9463799993792809]	[0.9505327445104533, 0.9683300291075624, 0.6951229456548329, 0.9726875536488669]
0.2449	4.5	720	0.1218	0.8479	0.8985	0.9641	[0.8961950935913541, 0.9301026110955253, 0.6193945084735495, 0.9457735746871611]	[0.9454922629374551, 0.9726557657238586, 0.7074187466486777, 0.9685915004384908]
0.1402	4.62	740	0.1205	0.8513	0.9089	0.9646	[0.8883404276098935, 0.9311559198709245, 0.6392782798591807, 0.9462383840511697]	[0.9563946865744879, 0.9610148600092611, 0.7421637743364691, 0.9758302728873157]
0.1993	4.75	760	0.1156	0.8513	0.9002	0.9650	[0.8973513861058572, 0.9315816840805964, 0.6294303742460324, 0.9468717796656311]	[0.9446992797434415, 0.9648566185002029, 0.7158152763466448, 0.9752565613203643]
0.1458	4.88	780	0.1162	0.8539	0.9091	0.9651	[0.8940502189002697, 0.932053937919347, 0.6428052076917813, 0.9468840079361927]	[0.9550781592275205, 0.9630270646289607, 0.7427543483793808, 0.9753759721895434]
0.0971	5.0	800	0.1208	0.8483	0.8963	0.9648	[0.896665866220625, 0.9313625624392157, 0.6184457649558176, 0.946788356429534]	[0.9488350594956452, 0.9690753342283529, 0.6948783145659281, 0.9725778146880949]

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Tokenizers 0.13.2