segformer-b5-finetuned-apple-dms-run6

This model is a fine-tuned version of nvidia/segformer-b5-finetuned-ade-640-640 on the AllanK24/apple-dms-materials dataset. It achieves the following results on the evaluation set:

Loss: 1.3095
Mean Iou: 0.4287
Mean Accuracy: 0.5210
Overall Accuracy: 0.8146
Accuracy Animal Skin: 0.6239
Iou Animal Skin: 0.3616
Accuracy Bone Teeth Horn: 0.0
Iou Bone Teeth Horn: 0.0
Accuracy Brickwork: 0.7311
Iou Brickwork: 0.5894
Accuracy Cardboard: 0.5931
Iou Cardboard: 0.4566
Accuracy Carpet Rug: 0.8736
Iou Carpet Rug: 0.7453
Accuracy Ceiling Tile: 0.8854
Iou Ceiling Tile: 0.7752
Accuracy Ceramic: 0.7765
Iou Ceramic: 0.6244
Accuracy Chalkboard Blackboard: 0.7451
Iou Chalkboard Blackboard: 0.6532
Accuracy Clutter: 0.0
Iou Clutter: 0.0
Accuracy Concrete: 0.5751
Iou Concrete: 0.4029
Accuracy Cork Corkboard: 0.0
Iou Cork Corkboard: 0.0
Accuracy Engineered Stone: 0.0
Iou Engineered Stone: 0.0
Accuracy Fabric Cloth: 0.8890
Iou Fabric Cloth: 0.7941
Accuracy Fiberglass Wool: 0.0
Iou Fiberglass Wool: 0.0
Accuracy Fire: 0.5013
Iou Fire: 0.4614
Accuracy Foliage: 0.9338
Iou Foliage: 0.8339
Accuracy Food: 0.9131
Iou Food: 0.7962
Accuracy Fur: 0.9065
Iou Fur: 0.8252
Accuracy Gemstone Quartz: 0.0
Iou Gemstone Quartz: 0.0
Accuracy Glass: 0.7477
Iou Glass: 0.6147
Accuracy Hair: 0.8372
Iou Hair: 0.7202
Accuracy Ice: 0.2066
Iou Ice: 0.1456
Accuracy Leather: 0.6290
Iou Leather: 0.5031
Accuracy Liquid Non-water: 0.0
Iou Liquid Non-water: 0.0
Accuracy Metal: 0.4857
Iou Metal: 0.3597
Accuracy Mirror: 0.5635
Iou Mirror: 0.4752
Accuracy Paint Plaster Enamel: 0.8706
Iou Paint Plaster Enamel: 0.7451
Accuracy Paper: 0.7491
Iou Paper: 0.5925
Accuracy Pearl: 0.0
Iou Pearl: 0.0
Accuracy Photograph Painting: 0.4782
Iou Photograph Painting: 0.3407
Accuracy Plastic Clear: 0.3854
Iou Plastic Clear: 0.2859
Accuracy Plastic Non-clear: 0.5527
Iou Plastic Non-clear: 0.3961
Accuracy Rubber Latex: 0.2813
Iou Rubber Latex: 0.2446
Accuracy Sand: 0.6207
Iou Sand: 0.4543
Accuracy Skin Lips: 0.8414
Iou Skin Lips: 0.7268
Accuracy Sky: 0.9760
Iou Sky: 0.9308
Accuracy Snow: 0.6967
Iou Snow: 0.5703
Accuracy Soap: 0.0
Iou Soap: 0.0
Accuracy Soil Mud: 0.6107
Iou Soil Mud: 0.4508
Accuracy Sponge: 0.0
Iou Sponge: 0.0
Accuracy Stone Natural: 0.7034
Iou Stone Natural: 0.5471
Accuracy Stone Polished: 0.3896
Iou Stone Polished: 0.2613
Accuracy Styrofoam: 0.0
Iou Styrofoam: 0.0
Accuracy Tile: 0.8096
Iou Tile: 0.6769
Accuracy Wallpaper: 0.5224
Iou Wallpaper: 0.3967
Accuracy Water: 0.8975
Iou Water: 0.8097
Accuracy Wax: 0.0
Iou Wax: 0.0
Accuracy Whiteboard: 0.8301
Iou Whiteboard: 0.7157
Accuracy Wicker: 0.5554
Iou Wicker: 0.4751
Accuracy Wood: 0.8548
Iou Wood: 0.7357
Accuracy Wood Tree: 0.4078
Iou Wood Tree: 0.3020
Accuracy Asphalt: 0.6427
Iou Asphalt: 0.4939

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: 32
eval_batch_size: 16
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 256
total_eval_batch_size: 128
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 0.1
num_epochs: 20
label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Animal Skin	Iou Animal Skin	Accuracy Brickwork	Iou Brickwork	Accuracy Cardboard	Iou Cardboard	Accuracy Carpet Rug	Iou Carpet Rug	Accuracy Ceiling Tile	Iou Ceiling Tile	Accuracy Ceramic	Iou Ceramic	Accuracy Chalkboard Blackboard	Iou Chalkboard Blackboard	Accuracy Concrete	Iou Concrete	Accuracy Fabric Cloth	Iou Fabric Cloth	Accuracy Fire	Iou Fire	Accuracy Foliage	Iou Foliage	Accuracy Food	Iou Food	Accuracy Fur	Iou Fur	Accuracy Glass	Iou Glass	Accuracy Hair	Iou Hair	Accuracy Ice	Iou Ice	Accuracy Leather	Iou Leather	Accuracy Metal	Iou Metal	Accuracy Mirror	Iou Mirror	Accuracy Paint Plaster Enamel	Iou Paint Plaster Enamel	Accuracy Paper	Iou Paper	Accuracy Photograph Painting	Iou Photograph Painting	Accuracy Plastic Clear	Iou Plastic Clear	Accuracy Plastic Non-clear	Iou Plastic Non-clear	Accuracy Rubber Latex	Iou Rubber Latex	Accuracy Sand	Iou Sand	Accuracy Skin Lips	Iou Skin Lips	Accuracy Sky	Iou Sky	Accuracy Snow	Iou Snow	Accuracy Soil Mud	Iou Soil Mud	Accuracy Stone Natural	Iou Stone Natural	Accuracy Stone Polished	Iou Stone Polished	Accuracy Tile	Iou Tile	Accuracy Wallpaper	Iou Wallpaper	Accuracy Water	Iou Water	Accuracy Whiteboard	Iou Whiteboard	Accuracy Wicker	Iou Wicker	Accuracy Wood	Iou Wood	Accuracy Wood Tree	Iou Wood Tree	Accuracy Asphalt	Iou Asphalt
3.1187	1.7045	150	2.0344	0.1283	0.1773	0.6652	0.0	0.0	0.0	0.0	0.0	0.0	0.8223	0.4870	0.6227	0.5929	0.0	0.0	0.0	0.0	0.0049	0.0049	0.8601	0.6524	0.0	0.0	0.9054	0.6439	0.0005	0.0005	0.0000	0.0000	0.5712	0.3713	0.0211	0.0210	0.0	0.0	0.0	0.0	0.0010	0.0009	0.0	0.0	0.8813	0.6449	0.4875	0.3236	0.0	0.0	0.0	0.0	0.0385	0.0342	0.0	0.0	0.0	0.0	0.6019	0.3939	0.9707	0.8688	0.0	0.0	0.0	0.0	0.0175	0.0173	0.0	0.0	0.6855	0.5055	0.0	0.0	0.8852	0.5844	0.0	0.0	0.0	0.0	0.8435	0.5221	0.0	0.0	0.0	0.0
1.6249	3.4091	300	1.4451	0.2622	0.3308	0.7654	0.0	0.0	0.6807	0.5119	0.0094	0.0094	0.8969	0.6791	0.8959	0.7556	0.6273	0.5174	0.0	0.0	0.5490	0.3291	0.8799	0.7403	0.0	0.0	0.9412	0.7916	0.9128	0.7080	0.8579	0.7202	0.7142	0.5561	0.7724	0.6438	0.0	0.0	0.2039	0.1958	0.3405	0.2523	0.4161	0.3413	0.8587	0.7149	0.7470	0.5032	0.0636	0.0617	0.0	0.0	0.3562	0.2606	0.0	0.0	0.0	0.0	0.7887	0.6437	0.9624	0.9042	0.0	0.0	0.0	0.0	0.7167	0.3347	0.0	0.0	0.7620	0.6158	0.4683	0.3998	0.9143	0.7486	0.0063	0.0060	0.0	0.0	0.8599	0.6896	0.0	0.0	0.0	0.0
1.2610	5.1136	450	1.3401	0.3198	0.4050	0.7874	0.0	0.0	0.7095	0.5598	0.6172	0.3573	0.8687	0.7238	0.9089	0.7498	0.7622	0.5726	0.0001	0.0001	0.5200	0.3170	0.8897	0.7686	0.0	0.0	0.9316	0.8179	0.9045	0.7712	0.8997	0.7677	0.7259	0.5857	0.7900	0.6708	0.0	0.0	0.5772	0.4526	0.4267	0.3098	0.5758	0.4144	0.8482	0.7290	0.7690	0.5382	0.3750	0.2898	0.1239	0.1191	0.5501	0.3549	0.0	0.0	0.0053	0.0053	0.7996	0.6707	0.9704	0.9174	0.0038	0.0038	0.6539	0.3717	0.6314	0.4673	0.0088	0.0088	0.8350	0.6301	0.6070	0.4454	0.9312	0.7759	0.6912	0.4475	0.0000	0.0000	0.8174	0.7087	0.0001	0.0001	0.3308	0.3054
1.1310	6.8182	600	1.3033	0.3882	0.4789	0.8015	0.1938	0.1429	0.7248	0.5619	0.6280	0.4557	0.8688	0.7285	0.9088	0.7564	0.7417	0.6006	0.7091	0.6166	0.6348	0.3979	0.8924	0.7786	0.0013	0.0013	0.9316	0.8266	0.9045	0.7942	0.9074	0.7669	0.7021	0.5909	0.8045	0.6862	0.0	0.0	0.5688	0.4728	0.5090	0.3429	0.5322	0.4337	0.8548	0.7311	0.7071	0.5686	0.4753	0.3359	0.3142	0.2575	0.4871	0.3620	0.0065	0.0065	0.4508	0.4005	0.8197	0.6873	0.9743	0.9229	0.6810	0.4751	0.7447	0.4311	0.6168	0.5116	0.3351	0.2275	0.8147	0.6619	0.7263	0.4456	0.8923	0.8055	0.8154	0.6674	0.4392	0.3960	0.8558	0.7203	0.2491	0.2215	0.4808	0.3964
1.0609	8.5227	750	1.3006	0.4057	0.5039	0.8069	0.6068	0.3137	0.7388	0.5703	0.6441	0.4513	0.8795	0.7267	0.8953	0.7657	0.7755	0.5993	0.7473	0.6409	0.6191	0.4037	0.8869	0.7842	0.0807	0.0806	0.9280	0.8337	0.9105	0.7910	0.9166	0.8078	0.7204	0.6015	0.8191	0.6965	0.0	0.0	0.5918	0.4874	0.5089	0.3486	0.5935	0.4482	0.8572	0.7404	0.7124	0.5790	0.4781	0.3310	0.3632	0.2724	0.5328	0.3821	0.1909	0.1828	0.4788	0.3792	0.8403	0.7030	0.9789	0.9265	0.6869	0.5106	0.6129	0.4191	0.6906	0.5211	0.3875	0.2435	0.7733	0.6648	0.6321	0.4459	0.9049	0.8035	0.8324	0.7015	0.5195	0.4411	0.8606	0.7282	0.3778	0.2808	0.6295	0.4882
1.0174	10.2273	900	1.3033	0.4135	0.5106	0.8108	0.5883	0.3106	0.7233	0.5864	0.6129	0.4570	0.8671	0.7409	0.9018	0.7627	0.7640	0.6112	0.7730	0.6285	0.5871	0.4002	0.9007	0.7858	0.2534	0.2478	0.9336	0.8307	0.9047	0.7983	0.9086	0.8165	0.7326	0.6087	0.8342	0.7018	0.0	0.0	0.6073	0.4919	0.4646	0.3468	0.5408	0.4584	0.8570	0.7437	0.7421	0.5806	0.5024	0.3427	0.3871	0.2817	0.5425	0.3851	0.2307	0.2091	0.6026	0.4309	0.8309	0.7115	0.9773	0.9260	0.7175	0.4860	0.5889	0.4469	0.6839	0.5284	0.3499	0.2465	0.8089	0.6719	0.6635	0.4831	0.8974	0.7981	0.8726	0.6933	0.5533	0.4632	0.8667	0.7328	0.3628	0.2765	0.6159	0.4798
0.9943	11.9318	1050	1.3105	0.4214	0.5106	0.8122	0.5926	0.3635	0.7157	0.5808	0.5808	0.4555	0.8642	0.7464	0.8962	0.7763	0.7929	0.6140	0.7443	0.6525	0.5933	0.3964	0.8866	0.7906	0.4114	0.3953	0.9344	0.8339	0.9080	0.8017	0.9132	0.8161	0.7141	0.6003	0.8360	0.7144	0.0462	0.0395	0.6108	0.4940	0.4808	0.3557	0.5647	0.4440	0.8724	0.7433	0.7409	0.5887	0.4505	0.3437	0.3484	0.2798	0.5198	0.3859	0.2478	0.2247	0.6618	0.5150	0.8428	0.7170	0.9731	0.9279	0.6720	0.5322	0.6288	0.4383	0.6772	0.5395	0.3743	0.2506	0.8212	0.6675	0.5465	0.4083	0.9032	0.8050	0.8126	0.7049	0.5582	0.4732	0.8692	0.7350	0.3723	0.2874	0.5698	0.4731
0.9779	13.6364	1200	1.3086	0.4252	0.5189	0.8131	0.6336	0.3625	0.7289	0.5750	0.6196	0.4679	0.8723	0.7445	0.8890	0.7719	0.7588	0.6182	0.7575	0.6412	0.6085	0.4043	0.8902	0.7919	0.4589	0.4320	0.9337	0.8336	0.9113	0.7951	0.9078	0.8276	0.7384	0.6102	0.8403	0.7137	0.1160	0.0941	0.6263	0.4983	0.4698	0.3527	0.5349	0.4609	0.8677	0.7437	0.7345	0.5946	0.4786	0.3411	0.3780	0.2892	0.5661	0.3982	0.2743	0.2414	0.6057	0.4577	0.8391	0.7230	0.9775	0.9301	0.7064	0.5605	0.6372	0.4502	0.7162	0.5415	0.3631	0.2542	0.8164	0.6723	0.5421	0.4034	0.8966	0.8082	0.8399	0.7171	0.5608	0.4717	0.8493	0.7340	0.4046	0.2968	0.6311	0.4884
0.9648	15.3409	1350	1.3081	0.4283	0.5214	0.8144	0.6365	0.3659	0.7246	0.5902	0.5987	0.4633	0.8682	0.7505	0.8805	0.7783	0.7720	0.6263	0.7396	0.6522	0.5955	0.4046	0.8930	0.7933	0.4904	0.4548	0.9330	0.8340	0.9122	0.7925	0.9024	0.8237	0.7277	0.6112	0.8348	0.7164	0.1273	0.0925	0.6293	0.5077	0.4916	0.3589	0.5530	0.4655	0.8717	0.7443	0.7471	0.5939	0.4654	0.3403	0.3762	0.2916	0.5500	0.3945	0.2829	0.2463	0.6236	0.4662	0.8439	0.7247	0.9760	0.9308	0.7220	0.5487	0.6332	0.4545	0.7122	0.5496	0.3964	0.2655	0.8019	0.6780	0.5831	0.4278	0.8956	0.8117	0.8054	0.7073	0.5708	0.4766	0.8548	0.7360	0.4141	0.2995	0.6775	0.5042
0.9584	17.0455	1500	1.3095	0.4287	0.5210	0.8146	0.6239	0.3616	0.7311	0.5894	0.5931	0.4566	0.8736	0.7453	0.8854	0.7752	0.7765	0.6244	0.7451	0.6532	0.5751	0.4029	0.8890	0.7941	0.5013	0.4614	0.9338	0.8339	0.9131	0.7962	0.9065	0.8252	0.7477	0.6147	0.8372	0.7202	0.2066	0.1456	0.6290	0.5031	0.4857	0.3597	0.5635	0.4752	0.8706	0.7451	0.7491	0.5925	0.4782	0.3407	0.3854	0.2859	0.5527	0.3961	0.2813	0.2446	0.6207	0.4543	0.8414	0.7268	0.9760	0.9308	0.6967	0.5703	0.6107	0.4508	0.7034	0.5471	0.3896	0.2613	0.8096	0.6769	0.5224	0.3967	0.8975	0.8097	0.8301	0.7157	0.5554	0.4751	0.8548	0.7357	0.4078	0.3020	0.6427	0.4939

Framework versions

Transformers 5.0.0
Pytorch 2.9.1+cu128
Datasets 4.5.0
Tokenizers 0.22.2

Downloads last month: 5

Safetensors

Model size

84.6M params

Tensor type

F32

Model tree for AllanK24/segformer-b5-finetuned-apple-dms-run6

Base model

nvidia/segformer-b5-finetuned-ade-640-640

Finetuned

(22)

this model