Hugging Face's logo

s246216
/
segformer-finetuned-sidewalk-10k-steps

Image Segmentation
Transformers
Safetensors
segformer
vision
Generated from Trainer
Inference Endpoints
Model card Files Community
s246216 commited on
Commit
00e613f
·
verified ·
1 Parent(s): 2597915

Model save

Browse files
Files changed (1) hide show
  1. README.md +43 -44
README.md CHANGED
@@ -3,8 +3,6 @@ library_name: transformers
3
  license: other
4
  base_model: nvidia/mit-b0
5
  tags:
6
- - image-segmentation
7
- - vision
8
  - generated_from_trainer
9
  model-index:
10
  - name: segformer-finetuned-sidewalk-10k-steps
@@ -16,81 +14,81 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # segformer-finetuned-sidewalk-10k-steps
18
 
19
- This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the segments/sidewalk-semantic dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.5370
22
- - Mean Iou: 0.3152
23
- - Mean Accuracy: 0.3836
24
- - Overall Accuracy: 0.8446
25
  - Accuracy Unlabeled: nan
26
- - Accuracy Flat-road: 0.7943
27
- - Accuracy Flat-sidewalk: 0.9502
28
- - Accuracy Flat-crosswalk: 0.7814
29
- - Accuracy Flat-cyclinglane: 0.7826
30
- - Accuracy Flat-parkingdriveway: 0.4144
31
  - Accuracy Flat-railtrack: nan
32
- - Accuracy Flat-curb: 0.5120
33
- - Accuracy Human-person: 0.8395
34
  - Accuracy Human-rider: 0.0
35
- - Accuracy Vehicle-car: 0.9422
36
  - Accuracy Vehicle-truck: 0.0
37
  - Accuracy Vehicle-bus: 0.0
38
  - Accuracy Vehicle-tramtrain: 0.0
39
  - Accuracy Vehicle-motorcycle: 0.0
40
- - Accuracy Vehicle-bicycle: 0.5202
41
  - Accuracy Vehicle-caravan: 0.0
42
  - Accuracy Vehicle-cartrailer: 0.0
43
- - Accuracy Construction-building: 0.8702
44
  - Accuracy Construction-door: 0.0
45
- - Accuracy Construction-wall: 0.6362
46
- - Accuracy Construction-fenceguardrail: 0.5326
47
  - Accuracy Construction-bridge: 0.0
48
  - Accuracy Construction-tunnel: nan
49
  - Accuracy Construction-stairs: 0.0
50
- - Accuracy Object-pole: 0.5517
51
  - Accuracy Object-trafficsign: 0.0
52
  - Accuracy Object-trafficlight: 0.0
53
- - Accuracy Nature-vegetation: 0.9156
54
- - Accuracy Nature-terrain: 0.8952
55
- - Accuracy Sky: 0.9616
56
  - Accuracy Void-ground: 0.0
57
- - Accuracy Void-dynamic: 0.0065
58
- - Accuracy Void-static: 0.3684
59
  - Accuracy Void-unclear: 0.0
60
  - Iou Unlabeled: nan
61
- - Iou Flat-road: 0.6934
62
- - Iou Flat-sidewalk: 0.8442
63
- - Iou Flat-crosswalk: 0.6564
64
- - Iou Flat-cyclinglane: 0.6281
65
- - Iou Flat-parkingdriveway: 0.3705
66
  - Iou Flat-railtrack: nan
67
- - Iou Flat-curb: 0.3551
68
- - Iou Human-person: 0.5645
69
  - Iou Human-rider: 0.0
70
- - Iou Vehicle-car: 0.7782
71
  - Iou Vehicle-truck: 0.0
72
  - Iou Vehicle-bus: 0.0
73
  - Iou Vehicle-tramtrain: 0.0
74
  - Iou Vehicle-motorcycle: 0.0
75
- - Iou Vehicle-bicycle: 0.3567
76
  - Iou Vehicle-caravan: 0.0
77
  - Iou Vehicle-cartrailer: 0.0
78
- - Iou Construction-building: 0.7275
79
  - Iou Construction-door: 0.0
80
- - Iou Construction-wall: 0.5010
81
- - Iou Construction-fenceguardrail: 0.4609
82
  - Iou Construction-bridge: 0.0
83
  - Iou Construction-tunnel: nan
84
  - Iou Construction-stairs: 0.0
85
- - Iou Object-pole: 0.3953
86
  - Iou Object-trafficsign: 0.0
87
  - Iou Object-trafficlight: 0.0
88
- - Iou Nature-vegetation: 0.8456
89
- - Iou Nature-terrain: 0.7594
90
- - Iou Sky: 0.9066
91
  - Iou Void-ground: 0.0
92
- - Iou Void-dynamic: 0.0064
93
- - Iou Void-static: 0.2362
94
  - Iou Void-unclear: 0.0
95
 
96
  ## Model description
@@ -216,7 +214,8 @@ The following hyperparameters were used during training:
216
  | 0.3414 | 92.0 | 9844 | 0.0 | 0.8142 | 0.0 | 0.3795 | 0.0 | nan | 0.5216 | 0.7561 | 0.4008 | 0.8221 | 0.3113 | nan | 0.7151 | 0.9353 | 0.6918 | 0.0 | 0.8487 | 0.9201 | 0.4441 | 0.0 | 0.0 | 0.9681 | nan | 0.6002 | 0.0 | 0.9382 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0308 | 0.0 | 0.3497 | 0.0 | 0.0 | 0.6695 | 0.0 | 0.2835 | 0.0 | nan | 0.3499 | 0.5082 | 0.2868 | 0.5564 | 0.2581 | nan | 0.6108 | 0.8425 | 0.4276 | 0.0 | 0.7369 | 0.8308 | 0.2881 | 0.0 | 0.0 | 0.9221 | nan | 0.3509 | 0.0 | 0.7752 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0291 | 0.0 | 0.2613 | 0.0 | 0.6469 | 0.3693 | 0.2899 | 0.8197 |
217
  | 0.3313 | 93.0 | 9951 | 0.0 | 0.8203 | 0.0 | 0.3932 | 0.0 | nan | 0.5104 | 0.7604 | 0.4028 | 0.8078 | 0.2869 | nan | 0.7376 | 0.9340 | 0.6909 | 0.0 | 0.8579 | 0.9290 | 0.4384 | 0.0 | 0.0 | 0.9618 | nan | 0.5972 | 0.0 | 0.9377 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0288 | 0.0 | 0.3603 | 0.0 | 0.0 | 0.6687 | 0.0 | 0.2859 | 0.0 | nan | 0.3457 | 0.5104 | 0.2881 | 0.5710 | 0.2444 | nan | 0.6219 | 0.8456 | 0.4244 | 0.0 | 0.7464 | 0.8328 | 0.2898 | 0.0 | 0.0 | 0.9233 | nan | 0.3474 | 0.0 | 0.7735 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0272 | 0.0 | 0.2651 | 0.0 | 0.6325 | 0.3695 | 0.2907 | 0.8223 |
218
  | 0.3479 | 93.4579 | 10000 | 0.0 | 0.8226 | 0.0 | 0.3548 | 0.0 | nan | 0.5230 | 0.7581 | 0.3967 | 0.8054 | 0.3003 | nan | 0.7238 | 0.9385 | 0.6829 | 0.0 | 0.8618 | 0.9102 | 0.4428 | 0.0 | 0.0 | 0.9672 | nan | 0.6017 | 0.0 | 0.9452 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0271 | 0.0 | 0.3487 | 0.0 | 0.0 | 0.6692 | 0.0 | 0.2751 | 0.0 | nan | 0.3465 | 0.5049 | 0.2854 | 0.5664 | 0.2522 | nan | 0.6174 | 0.8440 | 0.4224 | 0.0 | 0.7310 | 0.8289 | 0.2844 | 0.0 | 0.0 | 0.9220 | nan | 0.3442 | 0.0 | 0.7668 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0257 | 0.0 | 0.2604 | 0.0 | 0.6392 | 0.3681 | 0.2886 | 0.8204 |
219
- | 0.3479 | 93.4673 | 10001 | 0.5370 | 0.3152 | 0.3836 | 0.8446 | nan | 0.7943 | 0.9502 | 0.7814 | 0.7826 | 0.4144 | nan | 0.5120 | 0.8395 | 0.0 | 0.9422 | 0.0 | 0.0 | 0.0 | 0.0 | 0.5202 | 0.0 | 0.0 | 0.8702 | 0.0 | 0.6362 | 0.5326 | 0.0 | nan | 0.0 | 0.5517 | 0.0 | 0.0 | 0.9156 | 0.8952 | 0.9616 | 0.0 | 0.0065 | 0.3684 | 0.0 | nan | 0.6934 | 0.8442 | 0.6564 | 0.6281 | 0.3705 | nan | 0.3551 | 0.5645 | 0.0 | 0.7782 | 0.0 | 0.0 | 0.0 | 0.0 | 0.3567 | 0.0 | 0.0 | 0.7275 | 0.0 | 0.5010 | 0.4609 | 0.0 | nan | 0.0 | 0.3953 | 0.0 | 0.0 | 0.8456 | 0.7594 | 0.9066 | 0.0 | 0.0064 | 0.2362 | 0.0 |
 
220
 
221
 
222
  ### Framework versions
 
3
  license: other
4
  base_model: nvidia/mit-b0
5
  tags:
 
 
6
  - generated_from_trainer
7
  model-index:
8
  - name: segformer-finetuned-sidewalk-10k-steps
 
14
 
15
  # segformer-finetuned-sidewalk-10k-steps
16
 
17
+ This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.5870
20
+ - Mean Iou: 0.2962
21
+ - Mean Accuracy: 0.3656
22
+ - Overall Accuracy: 0.8302
23
  - Accuracy Unlabeled: nan
24
+ - Accuracy Flat-road: 0.7752
25
+ - Accuracy Flat-sidewalk: 0.9460
26
+ - Accuracy Flat-crosswalk: 0.8223
27
+ - Accuracy Flat-cyclinglane: 0.7550
28
+ - Accuracy Flat-parkingdriveway: 0.2805
29
  - Accuracy Flat-railtrack: nan
30
+ - Accuracy Flat-curb: 0.5378
31
+ - Accuracy Human-person: 0.6477
32
  - Accuracy Human-rider: 0.0
33
+ - Accuracy Vehicle-car: 0.9354
34
  - Accuracy Vehicle-truck: 0.0
35
  - Accuracy Vehicle-bus: 0.0
36
  - Accuracy Vehicle-tramtrain: 0.0
37
  - Accuracy Vehicle-motorcycle: 0.0
38
+ - Accuracy Vehicle-bicycle: 0.5094
39
  - Accuracy Vehicle-caravan: 0.0
40
  - Accuracy Vehicle-cartrailer: 0.0
41
+ - Accuracy Construction-building: 0.8737
42
  - Accuracy Construction-door: 0.0
43
+ - Accuracy Construction-wall: 0.5689
44
+ - Accuracy Construction-fenceguardrail: 0.5286
45
  - Accuracy Construction-bridge: 0.0
46
  - Accuracy Construction-tunnel: nan
47
  - Accuracy Construction-stairs: 0.0
48
+ - Accuracy Object-pole: 0.4050
49
  - Accuracy Object-trafficsign: 0.0
50
  - Accuracy Object-trafficlight: 0.0
51
+ - Accuracy Nature-vegetation: 0.9220
52
+ - Accuracy Nature-terrain: 0.8417
53
+ - Accuracy Sky: 0.9603
54
  - Accuracy Void-ground: 0.0
55
+ - Accuracy Void-dynamic: 0.0091
56
+ - Accuracy Void-static: 0.3817
57
  - Accuracy Void-unclear: 0.0
58
  - Iou Unlabeled: nan
59
+ - Iou Flat-road: 0.6953
60
+ - Iou Flat-sidewalk: 0.8257
61
+ - Iou Flat-crosswalk: 0.6744
62
+ - Iou Flat-cyclinglane: 0.5831
63
+ - Iou Flat-parkingdriveway: 0.2594
64
  - Iou Flat-railtrack: nan
65
+ - Iou Flat-curb: 0.3720
66
+ - Iou Human-person: 0.4758
67
  - Iou Human-rider: 0.0
68
+ - Iou Vehicle-car: 0.7439
69
  - Iou Vehicle-truck: 0.0
70
  - Iou Vehicle-bus: 0.0
71
  - Iou Vehicle-tramtrain: 0.0
72
  - Iou Vehicle-motorcycle: 0.0
73
+ - Iou Vehicle-bicycle: 0.2578
74
  - Iou Vehicle-caravan: 0.0
75
  - Iou Vehicle-cartrailer: 0.0
76
+ - Iou Construction-building: 0.6847
77
  - Iou Construction-door: 0.0
78
+ - Iou Construction-wall: 0.4535
79
+ - Iou Construction-fenceguardrail: 0.4437
80
  - Iou Construction-bridge: 0.0
81
  - Iou Construction-tunnel: nan
82
  - Iou Construction-stairs: 0.0
83
+ - Iou Object-pole: 0.2822
84
  - Iou Object-trafficsign: 0.0
85
  - Iou Object-trafficlight: 0.0
86
+ - Iou Nature-vegetation: 0.8452
87
+ - Iou Nature-terrain: 0.7081
88
+ - Iou Sky: 0.9161
89
  - Iou Void-ground: 0.0
90
+ - Iou Void-dynamic: 0.0090
91
+ - Iou Void-static: 0.2470
92
  - Iou Void-unclear: 0.0
93
 
94
  ## Model description
 
214
  | 0.3414 | 92.0 | 9844 | 0.0 | 0.8142 | 0.0 | 0.3795 | 0.0 | nan | 0.5216 | 0.7561 | 0.4008 | 0.8221 | 0.3113 | nan | 0.7151 | 0.9353 | 0.6918 | 0.0 | 0.8487 | 0.9201 | 0.4441 | 0.0 | 0.0 | 0.9681 | nan | 0.6002 | 0.0 | 0.9382 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0308 | 0.0 | 0.3497 | 0.0 | 0.0 | 0.6695 | 0.0 | 0.2835 | 0.0 | nan | 0.3499 | 0.5082 | 0.2868 | 0.5564 | 0.2581 | nan | 0.6108 | 0.8425 | 0.4276 | 0.0 | 0.7369 | 0.8308 | 0.2881 | 0.0 | 0.0 | 0.9221 | nan | 0.3509 | 0.0 | 0.7752 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0291 | 0.0 | 0.2613 | 0.0 | 0.6469 | 0.3693 | 0.2899 | 0.8197 |
215
  | 0.3313 | 93.0 | 9951 | 0.0 | 0.8203 | 0.0 | 0.3932 | 0.0 | nan | 0.5104 | 0.7604 | 0.4028 | 0.8078 | 0.2869 | nan | 0.7376 | 0.9340 | 0.6909 | 0.0 | 0.8579 | 0.9290 | 0.4384 | 0.0 | 0.0 | 0.9618 | nan | 0.5972 | 0.0 | 0.9377 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0288 | 0.0 | 0.3603 | 0.0 | 0.0 | 0.6687 | 0.0 | 0.2859 | 0.0 | nan | 0.3457 | 0.5104 | 0.2881 | 0.5710 | 0.2444 | nan | 0.6219 | 0.8456 | 0.4244 | 0.0 | 0.7464 | 0.8328 | 0.2898 | 0.0 | 0.0 | 0.9233 | nan | 0.3474 | 0.0 | 0.7735 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0272 | 0.0 | 0.2651 | 0.0 | 0.6325 | 0.3695 | 0.2907 | 0.8223 |
216
  | 0.3479 | 93.4579 | 10000 | 0.0 | 0.8226 | 0.0 | 0.3548 | 0.0 | nan | 0.5230 | 0.7581 | 0.3967 | 0.8054 | 0.3003 | nan | 0.7238 | 0.9385 | 0.6829 | 0.0 | 0.8618 | 0.9102 | 0.4428 | 0.0 | 0.0 | 0.9672 | nan | 0.6017 | 0.0 | 0.9452 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0271 | 0.0 | 0.3487 | 0.0 | 0.0 | 0.6692 | 0.0 | 0.2751 | 0.0 | nan | 0.3465 | 0.5049 | 0.2854 | 0.5664 | 0.2522 | nan | 0.6174 | 0.8440 | 0.4224 | 0.0 | 0.7310 | 0.8289 | 0.2844 | 0.0 | 0.0 | 0.9220 | nan | 0.3442 | 0.0 | 0.7668 | 0.0 | 0.0 | 0.0 | nan | 0.0 | 0.0257 | 0.0 | 0.2604 | 0.0 | 0.6392 | 0.3681 | 0.2886 | 0.8204 |
217
+ | 0.3479 | 93.4673 | 10001 | 0.0 | 0.8702 | 0.0 | 0.5326 | 0.0 | nan | 0.6362 | 0.7814 | 0.5120 | 0.7826 | 0.4144 | nan | 0.7943 | 0.9502 | 0.8395 | 0.0 | 0.8952 | 0.9156 | 0.5517 | 0.0 | 0.0 | 0.9616 | nan | 0.5202 | 0.0 | 0.9422 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0065 | 0.0 | 0.3684 | 0.0 | 0.0 | 0.7275 | 0.0 | 0.4609 | 0.0 | nan | 0.5010 | 0.6564 | 0.3551 | 0.6281 | 0.3705 | nan | 0.6934 | 0.8442 | 0.5645 | 0.0 | 0.7594 | 0.8456 | 0.3953 | 0.0 | 0.0 | 0.9066 | nan | 0.3567 | 0.0 | 0.7782 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0064 | 0.0 | 0.2362 | 0.0 | 0.5370 | 0.3836 | 0.3152 | 0.8446 |
218
+ | 0.3479 | 93.4766 | 10002 | 0.5870 | 0.2962 | 0.3656 | 0.8302 | nan | 0.7752 | 0.9460 | 0.8223 | 0.7550 | 0.2805 | nan | 0.5378 | 0.6477 | 0.0 | 0.9354 | 0.0 | 0.0 | 0.0 | 0.0 | 0.5094 | 0.0 | 0.0 | 0.8737 | 0.0 | 0.5689 | 0.5286 | 0.0 | nan | 0.0 | 0.4050 | 0.0 | 0.0 | 0.9220 | 0.8417 | 0.9603 | 0.0 | 0.0091 | 0.3817 | 0.0 | nan | 0.6953 | 0.8257 | 0.6744 | 0.5831 | 0.2594 | nan | 0.3720 | 0.4758 | 0.0 | 0.7439 | 0.0 | 0.0 | 0.0 | 0.0 | 0.2578 | 0.0 | 0.0 | 0.6847 | 0.0 | 0.4535 | 0.4437 | 0.0 | nan | 0.0 | 0.2822 | 0.0 | 0.0 | 0.8452 | 0.7081 | 0.9161 | 0.0 | 0.0090 | 0.2470 | 0.0 |
219
 
220
 
221
  ### Framework versions