Commit
•
f96bc6b
1
Parent(s):
da6900a
Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,156 +1,201 @@
|
|
|
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
|
|
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- generated_from_trainer
|
|
|
|
| 6 |
model-index:
|
| 7 |
- name: drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_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_14-batch-size16_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: 0.4672
|
| 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 |
-
| 0.
|
| 139 |
-
| 0.
|
| 140 |
-
| 0.
|
| 141 |
-
|
| 142 |
-
| 0.
|
| 143 |
-
| 0.
|
| 144 |
-
| 0.
|
| 145 |
-
| 0.
|
| 146 |
-
| 0.
|
| 147 |
-
| 0.
|
| 148 |
-
| 0.
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
|
| 156 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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_14-batch-size16_freeze_probs
|
| 11 |
model-index:
|
| 12 |
- name: drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs
|
| 13 |
results: []
|
| 14 |
---
|
| 15 |
|
| 16 |
+
drone-DinoVdeau-from-probs is a fine-tuned version of [drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs](https://huggingface.co/drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs). It achieves the following results on the test set:
|
|
|
|
| 17 |
|
|
|
|
| 18 |
|
|
|
|
|
|
|
| 19 |
- Loss: 0.4672
|
| 20 |
+
- RMSE: 0.1550
|
| 21 |
+
- MAE: 0.1155
|
| 22 |
+
- KL Divergence: 0.3295
|
| 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_14-batch-size16_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**: 94.0
|
| 66 |
+
- **Learning Rate**: 0.001
|
| 67 |
+
- **Train Batch Size**: 16
|
| 68 |
+
- **Eval Batch Size**: 16
|
| 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 | 0.4933677613735199 | 0.1294 | 0.1825 | 0.9903 | 0.001
|
| 98 |
+
2 | 0.47886165976524353 | 0.1262 | 0.1716 | 0.6847 | 0.001
|
| 99 |
+
3 | 0.4788369834423065 | 0.1271 | 0.1709 | 0.5498 | 0.001
|
| 100 |
+
4 | 0.47656363248825073 | 0.1278 | 0.1695 | 0.3131 | 0.001
|
| 101 |
+
5 | 0.4765072166919708 | 0.1277 | 0.1687 | 0.4013 | 0.001
|
| 102 |
+
6 | 0.47649845480918884 | 0.1243 | 0.1689 | 0.6370 | 0.001
|
| 103 |
+
7 | 0.47629299759864807 | 0.1292 | 0.1694 | 0.4314 | 0.001
|
| 104 |
+
8 | 0.4755041003227234 | 0.1267 | 0.1681 | 0.3379 | 0.001
|
| 105 |
+
9 | 0.47342246770858765 | 0.1250 | 0.1662 | 0.4916 | 0.001
|
| 106 |
+
10 | 0.47546806931495667 | 0.1277 | 0.1677 | 0.3348 | 0.001
|
| 107 |
+
11 | 0.4731104075908661 | 0.1255 | 0.1659 | 0.3524 | 0.001
|
| 108 |
+
12 | 0.47679492831230164 | 0.1306 | 0.1690 | 0.2383 | 0.001
|
| 109 |
+
13 | 0.4736888110637665 | 0.1223 | 0.1666 | 0.6968 | 0.001
|
| 110 |
+
14 | 0.4736703634262085 | 0.1254 | 0.1658 | 0.3983 | 0.001
|
| 111 |
+
15 | 0.4738818407058716 | 0.1248 | 0.1664 | 0.5620 | 0.001
|
| 112 |
+
16 | 0.47202879190444946 | 0.1231 | 0.1648 | 0.6049 | 0.001
|
| 113 |
+
17 | 0.47406336665153503 | 0.1265 | 0.1664 | 0.3072 | 0.001
|
| 114 |
+
18 | 0.4738321006298065 | 0.1253 | 0.1650 | 0.3350 | 0.001
|
| 115 |
+
19 | 0.476326048374176 | 0.1282 | 0.1672 | 0.2746 | 0.001
|
| 116 |
+
20 | 0.4755523204803467 | 0.1245 | 0.1670 | 0.5659 | 0.001
|
| 117 |
+
21 | 0.47340598702430725 | 0.1230 | 0.1662 | 0.6154 | 0.001
|
| 118 |
+
22 | 0.47443991899490356 | 0.1223 | 0.1677 | 0.7974 | 0.001
|
| 119 |
+
23 | 0.47205689549446106 | 0.1252 | 0.1639 | 0.2307 | 0.0001
|
| 120 |
+
24 | 0.4706146717071533 | 0.1217 | 0.1631 | 0.4219 | 0.0001
|
| 121 |
+
25 | 0.46876564621925354 | 0.1195 | 0.1612 | 0.5242 | 0.0001
|
| 122 |
+
26 | 0.46925392746925354 | 0.1190 | 0.1620 | 0.6159 | 0.0001
|
| 123 |
+
27 | 0.46849024295806885 | 0.1206 | 0.1607 | 0.4046 | 0.0001
|
| 124 |
+
28 | 0.46939656138420105 | 0.1220 | 0.1616 | 0.2860 | 0.0001
|
| 125 |
+
29 | 0.46892231702804565 | 0.1197 | 0.1614 | 0.4270 | 0.0001
|
| 126 |
+
30 | 0.46987923979759216 | 0.1225 | 0.1619 | 0.2625 | 0.0001
|
| 127 |
+
31 | 0.46842578053474426 | 0.1197 | 0.1607 | 0.3876 | 0.0001
|
| 128 |
+
32 | 0.46784707903862 | 0.1195 | 0.1600 | 0.4060 | 0.0001
|
| 129 |
+
33 | 0.46755874156951904 | 0.1193 | 0.1596 | 0.3688 | 0.0001
|
| 130 |
+
34 | 0.46766504645347595 | 0.1194 | 0.1600 | 0.3900 | 0.0001
|
| 131 |
+
35 | 0.4670174717903137 | 0.1189 | 0.1593 | 0.4282 | 0.0001
|
| 132 |
+
36 | 0.46679624915122986 | 0.1180 | 0.1591 | 0.4446 | 0.0001
|
| 133 |
+
37 | 0.46689239144325256 | 0.1185 | 0.1590 | 0.3942 | 0.0001
|
| 134 |
+
38 | 0.4664570987224579 | 0.1177 | 0.1588 | 0.4783 | 0.0001
|
| 135 |
+
39 | 0.4674011468887329 | 0.1190 | 0.1597 | 0.3868 | 0.0001
|
| 136 |
+
40 | 0.4677062928676605 | 0.1195 | 0.1599 | 0.3627 | 0.0001
|
| 137 |
+
41 | 0.46822381019592285 | 0.1211 | 0.1602 | 0.2655 | 0.0001
|
| 138 |
+
42 | 0.4664672613143921 | 0.1172 | 0.1589 | 0.5072 | 0.0001
|
| 139 |
+
43 | 0.46638762950897217 | 0.1177 | 0.1585 | 0.4306 | 0.0001
|
| 140 |
+
44 | 0.46708741784095764 | 0.1192 | 0.1594 | 0.4115 | 0.0001
|
| 141 |
+
45 | 0.46663177013397217 | 0.1171 | 0.1590 | 0.4417 | 0.0001
|
| 142 |
+
46 | 0.4663327634334564 | 0.1179 | 0.1585 | 0.3686 | 0.0001
|
| 143 |
+
47 | 0.46577510237693787 | 0.1172 | 0.1582 | 0.5090 | 0.0001
|
| 144 |
+
48 | 0.46634000539779663 | 0.1175 | 0.1589 | 0.5279 | 0.0001
|
| 145 |
+
49 | 0.46656596660614014 | 0.1183 | 0.1591 | 0.4497 | 0.0001
|
| 146 |
+
50 | 0.46755433082580566 | 0.1205 | 0.1595 | 0.2712 | 0.0001
|
| 147 |
+
51 | 0.46639156341552734 | 0.1172 | 0.1586 | 0.4008 | 0.0001
|
| 148 |
+
52 | 0.46591076254844666 | 0.1163 | 0.1583 | 0.4922 | 0.0001
|
| 149 |
+
53 | 0.4656851887702942 | 0.1178 | 0.1579 | 0.4274 | 0.0001
|
| 150 |
+
54 | 0.46629655361175537 | 0.1158 | 0.1585 | 0.4574 | 0.0001
|
| 151 |
+
55 | 0.46644341945648193 | 0.1189 | 0.1586 | 0.3486 | 0.0001
|
| 152 |
+
56 | 0.4661739766597748 | 0.1184 | 0.1584 | 0.3016 | 0.0001
|
| 153 |
+
57 | 0.46634721755981445 | 0.1181 | 0.1587 | 0.4163 | 0.0001
|
| 154 |
+
58 | 0.4673805236816406 | 0.1189 | 0.1593 | 0.3399 | 0.0001
|
| 155 |
+
59 | 0.4650005102157593 | 0.1170 | 0.1572 | 0.3686 | 0.0001
|
| 156 |
+
60 | 0.46599113941192627 | 0.1172 | 0.1584 | 0.4535 | 0.0001
|
| 157 |
+
61 | 0.4662201702594757 | 0.1179 | 0.1585 | 0.3751 | 0.0001
|
| 158 |
+
62 | 0.46614503860473633 | 0.1173 | 0.1583 | 0.3534 | 0.0001
|
| 159 |
+
63 | 0.4660026431083679 | 0.1163 | 0.1583 | 0.4048 | 0.0001
|
| 160 |
+
64 | 0.46711620688438416 | 0.1188 | 0.1588 | 0.2471 | 0.0001
|
| 161 |
+
65 | 0.46536803245544434 | 0.1166 | 0.1577 | 0.4526 | 0.0001
|
| 162 |
+
66 | 0.46573594212532043 | 0.1161 | 0.1582 | 0.5259 | 1e-05
|
| 163 |
+
67 | 0.4653942584991455 | 0.1173 | 0.1574 | 0.4252 | 1e-05
|
| 164 |
+
68 | 0.46487176418304443 | 0.1154 | 0.1572 | 0.4989 | 1e-05
|
| 165 |
+
69 | 0.465110719203949 | 0.1161 | 0.1570 | 0.4023 | 1e-05
|
| 166 |
+
70 | 0.466043084859848 | 0.1166 | 0.1576 | 0.4118 | 1e-05
|
| 167 |
+
71 | 0.46608996391296387 | 0.1177 | 0.1578 | 0.3075 | 1e-05
|
| 168 |
+
72 | 0.46582332253456116 | 0.1171 | 0.1580 | 0.3836 | 1e-05
|
| 169 |
+
73 | 0.4648771584033966 | 0.1154 | 0.1569 | 0.4544 | 1e-05
|
| 170 |
+
74 | 0.46466848254203796 | 0.1163 | 0.1567 | 0.4538 | 1e-05
|
| 171 |
+
75 | 0.46563470363616943 | 0.1166 | 0.1573 | 0.3348 | 1e-05
|
| 172 |
+
76 | 0.4647076725959778 | 0.1158 | 0.1571 | 0.4976 | 1e-05
|
| 173 |
+
77 | 0.4650570750236511 | 0.1163 | 0.1570 | 0.3934 | 1e-05
|
| 174 |
+
78 | 0.46495845913887024 | 0.1161 | 0.1571 | 0.3936 | 1e-05
|
| 175 |
+
79 | 0.46530288457870483 | 0.1159 | 0.1573 | 0.3759 | 1e-05
|
| 176 |
+
80 | 0.4647064805030823 | 0.1162 | 0.1567 | 0.4189 | 1e-05
|
| 177 |
+
81 | 0.46485888957977295 | 0.1158 | 0.1571 | 0.4751 | 1.0000000000000002e-06
|
| 178 |
+
82 | 0.4654049277305603 | 0.1161 | 0.1572 | 0.4335 | 1.0000000000000002e-06
|
| 179 |
+
83 | 0.46466442942619324 | 0.1161 | 0.1566 | 0.3906 | 1.0000000000000002e-06
|
| 180 |
+
84 | 0.46430692076683044 | 0.1157 | 0.1564 | 0.3855 | 1.0000000000000002e-06
|
| 181 |
+
85 | 0.46528080105781555 | 0.1173 | 0.1571 | 0.3372 | 1.0000000000000002e-06
|
| 182 |
+
86 | 0.46546733379364014 | 0.1184 | 0.1572 | 0.2969 | 1.0000000000000002e-06
|
| 183 |
+
87 | 0.4651782214641571 | 0.1173 | 0.1571 | 0.3572 | 1.0000000000000002e-06
|
| 184 |
+
88 | 0.4655611217021942 | 0.1151 | 0.1578 | 0.5179 | 1.0000000000000002e-06
|
| 185 |
+
89 | 0.4654468297958374 | 0.1177 | 0.1574 | 0.2948 | 1.0000000000000002e-06
|
| 186 |
+
90 | 0.4649873971939087 | 0.1167 | 0.1569 | 0.3427 | 1.0000000000000002e-06
|
| 187 |
+
91 | 0.4655340015888214 | 0.1173 | 0.1572 | 0.2790 | 1.0000000000000002e-07
|
| 188 |
+
92 | 0.4645555317401886 | 0.1153 | 0.1566 | 0.4153 | 1.0000000000000002e-07
|
| 189 |
+
93 | 0.4648568034172058 | 0.1153 | 0.1571 | 0.4664 | 1.0000000000000002e-07
|
| 190 |
+
94 | 0.4652610421180725 | 0.1159 | 0.1568 | 0.3859 | 1.0000000000000002e-07
|
| 191 |
+
|
| 192 |
+
|
| 193 |
+
---
|
| 194 |
+
|
| 195 |
+
# Framework Versions
|
| 196 |
+
|
| 197 |
+
- **Transformers**: 4.41.0
|
| 198 |
+
- **Pytorch**: 2.5.0+cu124
|
| 199 |
+
- **Datasets**: 3.0.2
|
| 200 |
+
- **Tokenizers**: 0.19.1
|
| 201 |
+
|