Commit
•
cadc474
1
Parent(s):
05745a4
Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,145 +1,190 @@
|
|
|
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
|
|
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- generated_from_trainer
|
|
|
|
| 6 |
model-index:
|
| 7 |
- name: drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs
|
| 8 |
results: []
|
| 9 |
---
|
| 10 |
|
| 11 |
-
|
| 12 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
|
| 14 |
-
# drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs
|
| 15 |
|
| 16 |
-
This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) on the None dataset.
|
| 17 |
-
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 0.4668
|
| 19 |
-
-
|
| 20 |
-
-
|
| 21 |
-
-
|
| 22 |
-
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
-
|
| 26 |
|
| 27 |
-
|
| 28 |
|
| 29 |
-
|
|
|
|
|
|
|
|
|
|
| 30 |
|
| 31 |
-
|
| 32 |
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
|
| 35 |
-
|
| 36 |
|
| 37 |
-
|
| 38 |
|
| 39 |
-
|
| 40 |
|
| 41 |
The following hyperparameters were used during training:
|
| 42 |
-
|
| 43 |
-
-
|
| 44 |
-
-
|
| 45 |
-
-
|
| 46 |
-
-
|
| 47 |
-
-
|
| 48 |
-
-
|
| 49 |
-
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
| 0.
|
| 76 |
-
| 0.
|
| 77 |
-
| 0.
|
| 78 |
-
| 0.
|
| 79 |
-
| 0.
|
| 80 |
-
| 0.
|
| 81 |
-
| 0.
|
| 82 |
-
| 0.
|
| 83 |
-
| 0.
|
| 84 |
-
| 0.
|
| 85 |
-
| 0.
|
| 86 |
-
| 0.
|
| 87 |
-
| 0.
|
| 88 |
-
| 0.
|
| 89 |
-
| 0.
|
| 90 |
-
| 0.
|
| 91 |
-
| 0.
|
| 92 |
-
| 0.
|
| 93 |
-
| 0.
|
| 94 |
-
| 0.
|
| 95 |
-
| 0.
|
| 96 |
-
| 0.
|
| 97 |
-
| 0.
|
| 98 |
-
| 0.
|
| 99 |
-
| 0.
|
| 100 |
-
| 0.
|
| 101 |
-
| 0.
|
| 102 |
-
| 0.
|
| 103 |
-
| 0.
|
| 104 |
-
| 0.
|
| 105 |
-
| 0.
|
| 106 |
-
| 0.
|
| 107 |
-
| 0.
|
| 108 |
-
| 0.
|
| 109 |
-
| 0.
|
| 110 |
-
| 0.
|
| 111 |
-
| 0.
|
| 112 |
-
| 0.
|
| 113 |
-
| 0.
|
| 114 |
-
| 0.
|
| 115 |
-
| 0.
|
| 116 |
-
| 0.
|
| 117 |
-
| 0.
|
| 118 |
-
| 0.
|
| 119 |
-
| 0.
|
| 120 |
-
| 0.
|
| 121 |
-
| 0.
|
| 122 |
-
| 0.
|
| 123 |
-
| 0.
|
| 124 |
-
| 0.
|
| 125 |
-
| 0.
|
| 126 |
-
| 0.
|
| 127 |
-
| 0.
|
| 128 |
-
| 0.
|
| 129 |
-
| 0.
|
| 130 |
-
| 0.
|
| 131 |
-
| 0.
|
| 132 |
-
| 0.
|
| 133 |
-
| 0.
|
| 134 |
-
| 0.
|
| 135 |
-
| 0.
|
| 136 |
-
| 0.
|
| 137 |
-
| 0.
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
---
|
| 3 |
+
language:
|
| 4 |
+
- eng
|
| 5 |
+
license: cc0-1.0
|
| 6 |
tags:
|
| 7 |
+
- multilabel-image-classification
|
| 8 |
+
- multilabel
|
| 9 |
- generated_from_trainer
|
| 10 |
+
base_model: drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs
|
| 11 |
model-index:
|
| 12 |
- name: drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs
|
| 13 |
results: []
|
| 14 |
---
|
| 15 |
|
| 16 |
+
drone-DinoVdeau-from-probs is a fine-tuned version of [drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs](https://huggingface.co/drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs). It achieves the following results on the test set:
|
|
|
|
| 17 |
|
|
|
|
| 18 |
|
|
|
|
|
|
|
| 19 |
- Loss: 0.4668
|
| 20 |
+
- RMSE: 0.1546
|
| 21 |
+
- MAE: 0.1143
|
| 22 |
+
- KL Divergence: 0.3931
|
| 23 |
+
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
# Model description
|
| 27 |
+
drone-DinoVdeau-from-probs is a model built on top of drone-DinoVdeau-from-probs-large-2024_11_15-batch-size32_freeze_probs model for underwater multilabel image classification.The classification head is a combination of linear, ReLU, batch normalization, and dropout layers.
|
| 28 |
|
| 29 |
+
The source code for training the model can be found in this [Git repository](https://github.com/SeatizenDOI/DinoVdeau).
|
| 30 |
|
| 31 |
+
- **Developed by:** [lombardata](https://huggingface.co/lombardata), credits to [César Leblanc](https://huggingface.co/CesarLeblanc) and [Victor Illien](https://huggingface.co/groderg)
|
| 32 |
|
| 33 |
+
---
|
| 34 |
+
|
| 35 |
+
# Intended uses & limitations
|
| 36 |
+
You can use the raw model for classify diverse marine species, encompassing coral morphotypes classes taken from the Global Coral Reef Monitoring Network (GCRMN), habitats classes and seagrass species.
|
| 37 |
|
| 38 |
+
---
|
| 39 |
|
| 40 |
+
# Training and evaluation data
|
| 41 |
+
Details on the estimated number of images for each class are given in the following table:
|
| 42 |
+
| Class | train | test | val | Total |
|
| 43 |
+
|:------------------------|--------:|-------:|------:|--------:|
|
| 44 |
+
| Acropore_branched | 1220 | 363 | 362 | 1945 |
|
| 45 |
+
| Acropore_digitised | 586 | 195 | 189 | 970 |
|
| 46 |
+
| Acropore_tabular | 308 | 133 | 119 | 560 |
|
| 47 |
+
| Algae | 4777 | 1372 | 1384 | 7533 |
|
| 48 |
+
| Dead_coral | 2513 | 671 | 693 | 3877 |
|
| 49 |
+
| Millepore | 136 | 55 | 59 | 250 |
|
| 50 |
+
| No_acropore_encrusting | 252 | 88 | 93 | 433 |
|
| 51 |
+
| No_acropore_massive | 2158 | 725 | 726 | 3609 |
|
| 52 |
+
| No_acropore_sub_massive | 2036 | 582 | 612 | 3230 |
|
| 53 |
+
| Rock | 5976 | 1941 | 1928 | 9845 |
|
| 54 |
+
| Rubble | 4851 | 1486 | 1474 | 7811 |
|
| 55 |
+
| Sand | 6155 | 2019 | 1990 | 10164 |
|
| 56 |
|
| 57 |
+
---
|
| 58 |
|
| 59 |
+
# Training procedure
|
| 60 |
|
| 61 |
+
## Training hyperparameters
|
| 62 |
|
| 63 |
The following hyperparameters were used during training:
|
| 64 |
+
|
| 65 |
+
- **Number of Epochs**: 83.0
|
| 66 |
+
- **Learning Rate**: 0.001
|
| 67 |
+
- **Train Batch Size**: 32
|
| 68 |
+
- **Eval Batch Size**: 32
|
| 69 |
+
- **Optimizer**: Adam
|
| 70 |
+
- **LR Scheduler Type**: ReduceLROnPlateau with a patience of 5 epochs and a factor of 0.1
|
| 71 |
+
- **Freeze Encoder**: Yes
|
| 72 |
+
- **Data Augmentation**: Yes
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
## Data Augmentation
|
| 76 |
+
Data were augmented using the following transformations :
|
| 77 |
+
|
| 78 |
+
Train Transforms
|
| 79 |
+
- **PreProcess**: No additional parameters
|
| 80 |
+
- **Resize**: probability=1.00
|
| 81 |
+
- **RandomHorizontalFlip**: probability=0.25
|
| 82 |
+
- **RandomVerticalFlip**: probability=0.25
|
| 83 |
+
- **ColorJiggle**: probability=0.25
|
| 84 |
+
- **RandomPerspective**: probability=0.25
|
| 85 |
+
- **Normalize**: probability=1.00
|
| 86 |
+
|
| 87 |
+
Val Transforms
|
| 88 |
+
- **PreProcess**: No additional parameters
|
| 89 |
+
- **Resize**: probability=1.00
|
| 90 |
+
- **Normalize**: probability=1.00
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
|
| 94 |
+
## Training results
|
| 95 |
+
Epoch | Validation Loss | MAE | RMSE | KL div | Learning Rate
|
| 96 |
+
--- | --- | --- | --- | --- | ---
|
| 97 |
+
1 | 0.4855400025844574 | 0.1364 | 0.1771 | 0.3101 | 0.001
|
| 98 |
+
2 | 0.47601452469825745 | 0.1247 | 0.1688 | 0.5077 | 0.001
|
| 99 |
+
3 | 0.4776814579963684 | 0.1230 | 0.1707 | 0.7896 | 0.001
|
| 100 |
+
4 | 0.47429159283638 | 0.1238 | 0.1672 | 0.4932 | 0.001
|
| 101 |
+
5 | 0.47457176446914673 | 0.1277 | 0.1669 | 0.2901 | 0.001
|
| 102 |
+
6 | 0.4749792814254761 | 0.1253 | 0.1674 | 0.4399 | 0.001
|
| 103 |
+
7 | 0.4744807779788971 | 0.1259 | 0.1671 | 0.4868 | 0.001
|
| 104 |
+
8 | 0.47424906492233276 | 0.1257 | 0.1672 | 0.3241 | 0.001
|
| 105 |
+
9 | 0.4729686379432678 | 0.1236 | 0.1658 | 0.4560 | 0.001
|
| 106 |
+
10 | 0.4750550389289856 | 0.1269 | 0.1679 | 0.2141 | 0.001
|
| 107 |
+
11 | 0.4733181595802307 | 0.1265 | 0.1663 | 0.2530 | 0.001
|
| 108 |
+
12 | 0.4758349061012268 | 0.1264 | 0.1684 | 0.3966 | 0.001
|
| 109 |
+
13 | 0.4722050428390503 | 0.1223 | 0.1650 | 0.6055 | 0.001
|
| 110 |
+
14 | 0.4747372567653656 | 0.1250 | 0.1666 | 0.4203 | 0.001
|
| 111 |
+
15 | 0.47325292229652405 | 0.1227 | 0.1662 | 0.6553 | 0.001
|
| 112 |
+
16 | 0.4734710156917572 | 0.1241 | 0.1656 | 0.3576 | 0.001
|
| 113 |
+
17 | 0.4721581041812897 | 0.1221 | 0.1643 | 0.4545 | 0.001
|
| 114 |
+
18 | 0.4723944365978241 | 0.1225 | 0.1647 | 0.4902 | 0.001
|
| 115 |
+
19 | 0.47289156913757324 | 0.1261 | 0.1650 | 0.3158 | 0.001
|
| 116 |
+
20 | 0.4697262644767761 | 0.1203 | 0.1623 | 0.4574 | 0.0001
|
| 117 |
+
21 | 0.46890661120414734 | 0.1197 | 0.1613 | 0.4569 | 0.0001
|
| 118 |
+
22 | 0.46905258297920227 | 0.1202 | 0.1617 | 0.4535 | 0.0001
|
| 119 |
+
23 | 0.4691086411476135 | 0.1210 | 0.1614 | 0.2971 | 0.0001
|
| 120 |
+
24 | 0.46915334463119507 | 0.1196 | 0.1616 | 0.3916 | 0.0001
|
| 121 |
+
25 | 0.4676876664161682 | 0.1181 | 0.1601 | 0.4516 | 0.0001
|
| 122 |
+
26 | 0.4679708480834961 | 0.1171 | 0.1605 | 0.6089 | 0.0001
|
| 123 |
+
27 | 0.4674595892429352 | 0.1182 | 0.1600 | 0.4741 | 0.0001
|
| 124 |
+
28 | 0.46810340881347656 | 0.1200 | 0.1606 | 0.3356 | 0.0001
|
| 125 |
+
29 | 0.4678303897380829 | 0.1181 | 0.1603 | 0.4330 | 0.0001
|
| 126 |
+
30 | 0.46800243854522705 | 0.1194 | 0.1602 | 0.3160 | 0.0001
|
| 127 |
+
31 | 0.4676785469055176 | 0.1179 | 0.1600 | 0.4190 | 0.0001
|
| 128 |
+
32 | 0.46752873063087463 | 0.1188 | 0.1598 | 0.3706 | 0.0001
|
| 129 |
+
33 | 0.46710190176963806 | 0.1181 | 0.1593 | 0.3504 | 0.0001
|
| 130 |
+
34 | 0.4670344293117523 | 0.1180 | 0.1594 | 0.3881 | 0.0001
|
| 131 |
+
35 | 0.4662601053714752 | 0.1166 | 0.1587 | 0.4398 | 0.0001
|
| 132 |
+
36 | 0.46657058596611023 | 0.1170 | 0.1587 | 0.4382 | 0.0001
|
| 133 |
+
37 | 0.4657588005065918 | 0.1163 | 0.1581 | 0.4330 | 0.0001
|
| 134 |
+
38 | 0.4659184217453003 | 0.1162 | 0.1583 | 0.4878 | 0.0001
|
| 135 |
+
39 | 0.46703553199768066 | 0.1178 | 0.1595 | 0.3791 | 0.0001
|
| 136 |
+
40 | 0.4664987027645111 | 0.1178 | 0.1588 | 0.3889 | 0.0001
|
| 137 |
+
41 | 0.46659526228904724 | 0.1184 | 0.1589 | 0.3222 | 0.0001
|
| 138 |
+
42 | 0.4655005633831024 | 0.1164 | 0.1579 | 0.4262 | 0.0001
|
| 139 |
+
43 | 0.4656265676021576 | 0.1162 | 0.1579 | 0.4611 | 0.0001
|
| 140 |
+
44 | 0.4655725955963135 | 0.1164 | 0.1580 | 0.4586 | 0.0001
|
| 141 |
+
45 | 0.46600833535194397 | 0.1158 | 0.1583 | 0.4368 | 0.0001
|
| 142 |
+
46 | 0.4660418927669525 | 0.1164 | 0.1582 | 0.4118 | 0.0001
|
| 143 |
+
47 | 0.46521857380867004 | 0.1154 | 0.1577 | 0.5424 | 0.0001
|
| 144 |
+
48 | 0.46598610281944275 | 0.1160 | 0.1586 | 0.5251 | 0.0001
|
| 145 |
+
49 | 0.46604350209236145 | 0.1161 | 0.1585 | 0.5007 | 0.0001
|
| 146 |
+
50 | 0.46660009026527405 | 0.1185 | 0.1586 | 0.2424 | 0.0001
|
| 147 |
+
51 | 0.4660661220550537 | 0.1162 | 0.1584 | 0.4171 | 0.0001
|
| 148 |
+
52 | 0.4649689793586731 | 0.1155 | 0.1575 | 0.4912 | 0.0001
|
| 149 |
+
53 | 0.4653578996658325 | 0.1169 | 0.1578 | 0.4030 | 0.0001
|
| 150 |
+
54 | 0.4660585820674896 | 0.1153 | 0.1585 | 0.4811 | 0.0001
|
| 151 |
+
55 | 0.46527624130249023 | 0.1167 | 0.1576 | 0.3774 | 0.0001
|
| 152 |
+
56 | 0.4654240906238556 | 0.1176 | 0.1575 | 0.3254 | 0.0001
|
| 153 |
+
57 | 0.4654492139816284 | 0.1162 | 0.1575 | 0.3649 | 0.0001
|
| 154 |
+
58 | 0.46654412150382996 | 0.1166 | 0.1584 | 0.4075 | 0.0001
|
| 155 |
+
59 | 0.465238481760025 | 0.1157 | 0.1575 | 0.4202 | 1e-05
|
| 156 |
+
60 | 0.46530231833457947 | 0.1157 | 0.1571 | 0.4084 | 1e-05
|
| 157 |
+
61 | 0.4653523564338684 | 0.1153 | 0.1573 | 0.4497 | 1e-05
|
| 158 |
+
62 | 0.46477487683296204 | 0.1153 | 0.1568 | 0.4112 | 1e-05
|
| 159 |
+
63 | 0.46481335163116455 | 0.1152 | 0.1567 | 0.3748 | 1e-05
|
| 160 |
+
64 | 0.46523070335388184 | 0.1162 | 0.1571 | 0.3044 | 1e-05
|
| 161 |
+
65 | 0.46484872698783875 | 0.1153 | 0.1569 | 0.4685 | 1e-05
|
| 162 |
+
66 | 0.46500927209854126 | 0.1148 | 0.1573 | 0.5087 | 1e-05
|
| 163 |
+
67 | 0.4645930230617523 | 0.1155 | 0.1568 | 0.4274 | 1e-05
|
| 164 |
+
68 | 0.46456360816955566 | 0.1144 | 0.1566 | 0.4969 | 1e-05
|
| 165 |
+
69 | 0.464430034160614 | 0.1145 | 0.1564 | 0.4480 | 1e-05
|
| 166 |
+
70 | 0.4648461937904358 | 0.1150 | 0.1567 | 0.4291 | 1e-05
|
| 167 |
+
71 | 0.4645022749900818 | 0.1156 | 0.1565 | 0.3797 | 1e-05
|
| 168 |
+
72 | 0.46473589539527893 | 0.1150 | 0.1569 | 0.4280 | 1e-05
|
| 169 |
+
73 | 0.46414923667907715 | 0.1142 | 0.1563 | 0.4592 | 1e-05
|
| 170 |
+
74 | 0.4641610085964203 | 0.1151 | 0.1564 | 0.4321 | 1e-05
|
| 171 |
+
75 | 0.4644509255886078 | 0.1152 | 0.1565 | 0.3843 | 1e-05
|
| 172 |
+
76 | 0.4646488130092621 | 0.1147 | 0.1569 | 0.5216 | 1e-05
|
| 173 |
+
77 | 0.46475714445114136 | 0.1152 | 0.1569 | 0.4094 | 1e-05
|
| 174 |
+
78 | 0.46428272128105164 | 0.1149 | 0.1564 | 0.4399 | 1e-05
|
| 175 |
+
79 | 0.4645934998989105 | 0.1147 | 0.1567 | 0.4178 | 1e-05
|
| 176 |
+
80 | 0.46436014771461487 | 0.1150 | 0.1564 | 0.4373 | 1.0000000000000002e-06
|
| 177 |
+
81 | 0.46448636054992676 | 0.1151 | 0.1567 | 0.4701 | 1.0000000000000002e-06
|
| 178 |
+
82 | 0.4644375145435333 | 0.1146 | 0.1565 | 0.4601 | 1.0000000000000002e-06
|
| 179 |
+
83 | 0.46457409858703613 | 0.1147 | 0.1567 | 0.4511 | 1.0000000000000002e-06
|
| 180 |
+
|
| 181 |
+
|
| 182 |
+
---
|
| 183 |
+
|
| 184 |
+
# Framework Versions
|
| 185 |
+
|
| 186 |
+
- **Transformers**: 4.41.0
|
| 187 |
+
- **Pytorch**: 2.5.0+cu124
|
| 188 |
+
- **Datasets**: 3.0.2
|
| 189 |
+
- **Tokenizers**: 0.19.1
|
| 190 |
+
|