nvidia-mit-b3_finetune_sidewalk-semantic

This model is a fine-tuned version of nvidia/mit-b3 on the segments/sidewalk-semantic dataset. It achieves the following results on the evaluation set:

• Loss: 0.7270
• Mean Iou: 0.2391
• Mean Accuracy: 0.2831
• Overall Accuracy: 0.8089
• Acc Unlabeled: nan
• Acc Flat-road: 0.8462
• Acc Flat-sidewalk: 0.9578
• Acc Flat-crosswalk: 0.0
• Acc Flat-cyclinglane: 0.4756
• Acc Flat-parkingdriveway: 0.2145
• Acc Flat-railtrack: nan
• Acc Flat-curb: 0.3585
• Acc Human-person: 0.2813
• Acc Human-rider: 0.0
• Acc Vehicle-car: 0.9514
• Acc Vehicle-truck: 0.0
• Acc Vehicle-bus: 0.0
• Acc Vehicle-tramtrain: 0.0
• Acc Vehicle-motorcycle: 0.0
• Acc Vehicle-bicycle: 0.2160
• Acc Vehicle-caravan: 0.0
• Acc Vehicle-cartrailer: 0.0
• Acc Construction-building: 0.9204
• Acc Construction-door: 0.0
• Acc Construction-wall: 0.2528
• Acc Construction-fenceguardrail: 0.3314
• Acc Construction-bridge: 0.0
• Acc Construction-tunnel: nan
• Acc Construction-stairs: 0.0
• Acc Object-pole: 0.3098
• Acc Object-trafficsign: 0.0
• Acc Object-trafficlight: 0.0
• Acc Nature-vegetation: 0.9375
• Acc Nature-terrain: 0.8768
• Acc Sky: 0.9406
• Acc Void-ground: 0.0
• Acc Void-dynamic: 0.0
• Acc Void-static: 0.1895
• Acc Void-unclear: 0.0
• Iou Unlabeled: nan
• Iou Flat-road: 0.6618
• Iou Flat-sidewalk: 0.8235
• Iou Flat-crosswalk: 0.0
• Iou Flat-cyclinglane: 0.4373
• Iou Flat-parkingdriveway: 0.1762
• Iou Flat-railtrack: nan
• Iou Flat-curb: 0.2784
• Iou Human-person: 0.2378
• Iou Human-rider: 0.0
• Iou Vehicle-car: 0.7833
• Iou Vehicle-truck: 0.0
• Iou Vehicle-bus: 0.0
• Iou Vehicle-tramtrain: 0.0
• Iou Vehicle-motorcycle: 0.0
• Iou Vehicle-bicycle: 0.2063
• Iou Vehicle-caravan: 0.0
• Iou Vehicle-cartrailer: 0.0
• Iou Construction-building: 0.6478
• Iou Construction-door: 0.0
• Iou Construction-wall: 0.2196
• Iou Construction-fenceguardrail: 0.3081
• Iou Construction-bridge: 0.0
• Iou Construction-tunnel: nan
• Iou Construction-stairs: 0.0
• Iou Object-pole: 0.2490
• Iou Object-trafficsign: 0.0
• Iou Object-trafficlight: 0.0
• Iou Nature-vegetation: 0.8362
• Iou Nature-terrain: 0.7356
• Iou Sky: 0.8856
• Iou Void-ground: 0.0
• Iou Void-dynamic: 0.0
• Iou Void-static: 0.1633
• Iou Void-unclear: 0.0

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: 5e-05
• train_batch_size: 3
• eval_batch_size: 3
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Acc Unlabeled	Acc Flat-road	Acc Flat-sidewalk	Acc Flat-crosswalk	Acc Flat-cyclinglane	Acc Flat-parkingdriveway	Acc Flat-railtrack	Acc Flat-curb	Acc Human-person	Acc Human-rider	Acc Vehicle-car	Acc Vehicle-truck	Acc Vehicle-bus	Acc Vehicle-tramtrain	Acc Vehicle-motorcycle	Acc Vehicle-bicycle	Acc Vehicle-caravan	Acc Vehicle-cartrailer	Acc Construction-building	Acc Construction-door	Acc Construction-wall	Acc Construction-fenceguardrail	Acc Construction-bridge	Acc Construction-tunnel	Acc Construction-stairs	Acc Object-pole	Acc Object-trafficsign	Acc Object-trafficlight	Acc Nature-vegetation	Acc Nature-terrain	Acc Sky	Acc Void-ground	Acc Void-dynamic	Acc Void-static	Acc Void-unclear	Iou Unlabeled	Iou Flat-road	Iou Flat-sidewalk	Iou Flat-crosswalk	Iou Flat-cyclinglane	Iou Flat-parkingdriveway	Iou Flat-railtrack	Iou Flat-curb	Iou Human-person	Iou Human-rider	Iou Vehicle-car	Iou Vehicle-truck	Iou Vehicle-bus	Iou Vehicle-tramtrain	Iou Vehicle-motorcycle	Iou Vehicle-bicycle	Iou Vehicle-caravan	Iou Vehicle-cartrailer	Iou Construction-building	Iou Construction-door	Iou Construction-wall	Iou Construction-fenceguardrail	Iou Construction-bridge	Iou Construction-tunnel	Iou Construction-stairs	Iou Object-pole	Iou Object-trafficsign	Iou Object-trafficlight	Iou Nature-vegetation	Iou Nature-terrain	Iou Sky	Iou Void-ground	Iou Void-dynamic	Iou Void-static	Iou Void-unclear
1.2696	0.14	33	1.3198	0.1366	0.1799	0.7212	nan	0.7268	0.9591	0.0	0.0	0.0054	nan	0.0	0.0	0.0	0.9144	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.8191	0.0	0.0032	0.0002	0.0	nan	0.0	0.0	0.0	0.0	0.9455	0.5741	0.8065	0.0	0.0	0.0014	0.0	nan	0.5092	0.7200	0.0	0.0	0.0052	nan	0.0	0.0	0.0	0.7334	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.5417	0.0	0.0032	0.0002	0.0	0.0	0.0	0.0	0.0	0.0	0.7237	0.4871	0.7824	0.0	0.0	0.0014	0.0
1.196	0.28	66	0.9744	0.1660	0.2092	0.7621	nan	0.8403	0.9446	0.0	0.2749	0.0164	nan	0.0000	0.0	0.0	0.9510	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.8910	0.0	0.0108	0.0620	0.0	nan	0.0	0.0256	0.0	0.0	0.9337	0.8260	0.9169	0.0	0.0	0.0018	0.0	nan	0.5760	0.7791	0.0	0.2646	0.0159	nan	0.0000	0.0	0.0	0.7093	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.5887	0.0	0.0107	0.0618	0.0	nan	0.0	0.0253	0.0	0.0	0.7845	0.6386	0.8554	0.0	0.0	0.0018	0.0
1.3009	0.42	99	0.8567	0.1940	0.2355	0.7789	nan	0.8392	0.9534	0.0	0.3587	0.0450	nan	0.0226	0.0861	0.0	0.9475	0.0	0.0	0.0	0.0	0.0085	0.0	0.0	0.9180	0.0	0.2409	0.2594	0.0	nan	0.0	0.1095	0.0	0.0	0.9179	0.8459	0.9262	0.0	0.0	0.0561	0.0	nan	0.5901	0.7918	0.0	0.3215	0.0422	nan	0.0221	0.0815	0.0	0.7401	0.0	0.0	0.0	0.0	0.0085	0.0	0.0	0.6252	0.0	0.2113	0.2513	0.0	nan	0.0	0.1028	0.0	0.0	0.8096	0.6881	0.8659	0.0	0.0	0.0543	0.0
0.8802	0.56	132	0.8121	0.2111	0.2533	0.7870	nan	0.7684	0.9678	0.0	0.5098	0.0480	nan	0.1309	0.1147	0.0	0.9545	0.0	0.0	0.0	0.0	0.1207	0.0	0.0	0.8997	0.0	0.3134	0.2572	0.0	nan	0.0	0.2145	0.0	0.0	0.9277	0.8730	0.9489	0.0	0.0	0.0552	0.0	nan	0.6078	0.7799	0.0	0.4300	0.0452	nan	0.1190	0.1086	0.0	0.7451	0.0	0.0	0.0	0.0	0.1171	0.0	0.0	0.6525	0.0	0.2578	0.2467	0.0	nan	0.0	0.1851	0.0	0.0	0.8146	0.7176	0.8735	0.0	0.0	0.0531	0.0
0.6118	0.71	165	0.7667	0.2267	0.2697	0.7976	nan	0.8872	0.9438	0.0	0.4101	0.1549	nan	0.1201	0.2090	0.0	0.9404	0.0	0.0	0.0	0.0	0.1452	0.0	0.0	0.8688	0.0	0.3950	0.3483	0.0	nan	0.0	0.2586	0.0	0.0	0.9422	0.8637	0.9486	0.0	0.0	0.1931	0.0	nan	0.6038	0.8210	0.0	0.3802	0.1356	nan	0.1094	0.1871	0.0	0.7863	0.0	0.0	0.0	0.0	0.1413	0.0	0.0	0.6519	0.0	0.2892	0.3200	0.0	nan	0.0	0.2204	0.0	0.0	0.8268	0.7340	0.8818	0.0	0.0	0.1644	0.0
1.3374	0.85	198	0.7325	0.2339	0.2768	0.8064	nan	0.8797	0.9501	0.0	0.4559	0.2181	nan	0.2268	0.2519	0.0	0.9457	0.0	0.0	0.0	0.0	0.2066	0.0	0.0	0.9112	0.0	0.2305	0.3397	0.0	nan	0.0	0.2854	0.0	0.0	0.9403	0.8881	0.9445	0.0	0.0	0.1843	0.0	nan	0.6533	0.8263	0.0	0.4195	0.1737	nan	0.1954	0.2214	0.0	0.7893	0.0	0.0	0.0	0.0	0.1979	0.0	0.0	0.6490	0.0	0.2057	0.3186	0.0	nan	0.0	0.2353	0.0	0.0	0.8301	0.7274	0.8832	0.0	0.0	0.1599	0.0
0.9023	0.99	231	0.7270	0.2391	0.2831	0.8089	nan	0.8462	0.9578	0.0	0.4756	0.2145	nan	0.3585	0.2813	0.0	0.9514	0.0	0.0	0.0	0.0	0.2160	0.0	0.0	0.9204	0.0	0.2528	0.3314	0.0	nan	0.0	0.3098	0.0	0.0	0.9375	0.8768	0.9406	0.0	0.0	0.1895	0.0	nan	0.6618	0.8235	0.0	0.4373	0.1762	nan	0.2784	0.2378	0.0	0.7833	0.0	0.0	0.0	0.0	0.2063	0.0	0.0	0.6478	0.0	0.2196	0.3081	0.0	nan	0.0	0.2490	0.0	0.0	0.8362	0.7356	0.8856	0.0	0.0	0.1633	0.0

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu117
• Datasets 2.20.0
• Tokenizers 0.13.3