File size: 13,929 Bytes
06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 71af6cd 06e48e5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 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 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 |
---
language:
- eng
license: wtfpl
tags:
- multilabel-image-classification
- multilabel
- generated_from_trainer
base_model: facebook/dinov2-large
model-index:
- name: Joseph-large-2024_09_16-batch-size32_epochs150_freeze
results: []
---
DinoVd'eau is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large). It achieves the following results on the test set:
- Loss: 0.1207
- F1 Micro: 0.8214
- F1 Macro: 0.7191
- Roc Auc: 0.8814
- Accuracy: 0.3118
---
# Model description
DinoVd'eau is a model built on top of dinov2 model for underwater multilabel image classification.The classification head is a combination of linear, ReLU, batch normalization, and dropout layers.
The source code for training the model can be found in this [Git repository](https://github.com/SeatizenDOI/DinoVdeau).
- **Developed by:** [lombardata](https://huggingface.co/lombardata), credits to [César Leblanc](https://huggingface.co/CesarLeblanc) and [Victor Illien](https://huggingface.co/groderg)
---
# Intended uses & limitations
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.
---
# Training and evaluation data
Details on the number of images for each class are given in the following table:
| Class | train | val | test | Total |
|:-------------------------|--------:|------:|-------:|--------:|
| Acropore_branched | 1469 | 464 | 475 | 2408 |
| Acropore_digitised | 568 | 160 | 160 | 888 |
| Acropore_sub_massive | 150 | 50 | 43 | 243 |
| Acropore_tabular | 999 | 297 | 293 | 1589 |
| Algae_assembly | 2546 | 847 | 845 | 4238 |
| Algae_drawn_up | 367 | 126 | 127 | 620 |
| Algae_limestone | 1652 | 557 | 563 | 2772 |
| Algae_sodding | 3148 | 984 | 985 | 5117 |
| Atra/Leucospilota | 1084 | 348 | 360 | 1792 |
| Bleached_coral | 219 | 71 | 70 | 360 |
| Blurred | 191 | 67 | 62 | 320 |
| Dead_coral | 1979 | 642 | 643 | 3264 |
| Fish | 2018 | 656 | 647 | 3321 |
| Homo_sapiens | 161 | 62 | 59 | 282 |
| Human_object | 157 | 58 | 55 | 270 |
| Living_coral | 406 | 154 | 141 | 701 |
| Millepore | 385 | 127 | 125 | 637 |
| No_acropore_encrusting | 441 | 130 | 154 | 725 |
| No_acropore_foliaceous | 204 | 36 | 46 | 286 |
| No_acropore_massive | 1031 | 336 | 338 | 1705 |
| No_acropore_solitary | 202 | 53 | 48 | 303 |
| No_acropore_sub_massive | 1401 | 433 | 422 | 2256 |
| Rock | 4489 | 1495 | 1473 | 7457 |
| Rubble | 3092 | 1030 | 1001 | 5123 |
| Sand | 5842 | 1939 | 1938 | 9719 |
| Sea_cucumber | 1408 | 439 | 447 | 2294 |
| Sea_urchins | 327 | 107 | 111 | 545 |
| Sponge | 269 | 96 | 105 | 470 |
| Syringodium_isoetifolium | 1212 | 392 | 391 | 1995 |
| Thalassodendron_ciliatum | 782 | 261 | 260 | 1303 |
| Useless | 579 | 193 | 193 | 965 |
---
# Training procedure
## Training hyperparameters
The following hyperparameters were used during training:
- **Number of Epochs**: 150
- **Learning Rate**: 0.001
- **Train Batch Size**: 32
- **Eval Batch Size**: 32
- **Optimizer**: Adam
- **LR Scheduler Type**: ReduceLROnPlateau with a patience of 5 epochs and a factor of 0.1
- **Freeze Encoder**: Yes
- **Data Augmentation**: Yes
## Data Augmentation
Data were augmented using the following transformations :
Train Transforms
- **PreProcess**: No additional parameters
- **Resize**: probability=1.00
- **RandomHorizontalFlip**: probability=0.25
- **RandomVerticalFlip**: probability=0.25
- **ColorJiggle**: probability=0.25
- **RandomPerspective**: probability=0.25
- **Normalize**: probability=1.00
Val Transforms
- **PreProcess**: No additional parameters
- **Resize**: probability=1.00
- **Normalize**: probability=1.00
## Training results
Epoch | Validation Loss | Accuracy | F1 Macro | F1 Micro | Learning Rate
--- | --- | --- | --- | --- | ---
1 | 0.17758780717849731 | 0.21656271656271656 | 0.7477812526413659 | 0.5384503258991854 | 0.001
2 | 0.153945192694664 | 0.24532224532224534 | 0.7697450182129848 | 0.5760774961516321 | 0.001
3 | 0.14735348522663116 | 0.2515592515592516 | 0.7744839226208509 | 0.6098114408992151 | 0.001
4 | 0.14645476639270782 | 0.25363825363825365 | 0.7738915615654661 | 0.6213514572326843 | 0.001
5 | 0.14515458047389984 | 0.25017325017325015 | 0.78146492434663 | 0.6353051230272125 | 0.001
6 | 0.1445809006690979 | 0.2577962577962578 | 0.781259480778399 | 0.6141782571643486 | 0.001
7 | 0.14445114135742188 | 0.26195426195426197 | 0.7800943800943801 | 0.6232727577909734 | 0.001
8 | 0.14366209506988525 | 0.25848925848925847 | 0.7879197465681098 | 0.6339480584029394 | 0.001
9 | 0.1447097659111023 | 0.2577962577962578 | 0.785476860138072 | 0.6442804243684905 | 0.001
10 | 0.1538563072681427 | 0.2442827442827443 | 0.7683399403144626 | 0.6149084687726756 | 0.001
11 | 0.1389196366071701 | 0.26334026334026334 | 0.7896514859952961 | 0.6334773464226039 | 0.001
12 | 0.1395249217748642 | 0.26403326403326405 | 0.7908438442264407 | 0.6406158966836866 | 0.001
13 | 0.1390257179737091 | 0.26507276507276506 | 0.7893533497260687 | 0.6557265830014797 | 0.001
14 | 0.13910652697086334 | 0.2623007623007623 | 0.787792943600309 | 0.640540413256037 | 0.001
15 | 0.13990363478660583 | 0.253984753984754 | 0.7885381419454319 | 0.6406412255611948 | 0.001
16 | 0.13938209414482117 | 0.2668052668052668 | 0.7847859161051945 | 0.6374513053376879 | 0.001
17 | 0.15936270356178284 | 0.24185724185724186 | 0.7857319587628866 | 0.6424904129432089 | 0.001
18 | 0.13188092410564423 | 0.27546777546777546 | 0.8036556603773585 | 0.6768028620378452 | 0.0001
19 | 0.13244545459747314 | 0.27893277893277896 | 0.8038422649140546 | 0.6715138701269487 | 0.0001
20 | 0.1306440383195877 | 0.27893277893277896 | 0.8066104665720725 | 0.6733647561041333 | 0.0001
21 | 0.1302667111158371 | 0.2817047817047817 | 0.8037271837637748 | 0.6728395801753237 | 0.0001
22 | 0.12870918214321136 | 0.2841302841302841 | 0.8074214632089395 | 0.6735047356746011 | 0.0001
23 | 0.1287251114845276 | 0.2841302841302841 | 0.8058198574902932 | 0.678520497542563 | 0.0001
24 | 0.1279863715171814 | 0.2869022869022869 | 0.8057504997660669 | 0.6840871439155845 | 0.0001
25 | 0.127402663230896 | 0.28586278586278585 | 0.8074392712550608 | 0.6787317976982782 | 0.0001
26 | 0.12828372418880463 | 0.28586278586278585 | 0.8063818050664064 | 0.6740298841901063 | 0.0001
27 | 0.12681305408477783 | 0.2882882882882883 | 0.8110456615281781 | 0.68897744745899 | 0.0001
28 | 0.12666279077529907 | 0.28932778932778935 | 0.8099940913311386 | 0.6812786729949134 | 0.0001
29 | 0.12675043940544128 | 0.29175329175329173 | 0.8081058020477816 | 0.6881122302734826 | 0.0001
30 | 0.12635387480258942 | 0.2927927927927928 | 0.8108657880239013 | 0.6872571297964245 | 0.0001
31 | 0.1258317530155182 | 0.29140679140679143 | 0.8089332139965051 | 0.6823767206574823 | 0.0001
32 | 0.1260402798652649 | 0.29313929313929316 | 0.8112645318336341 | 0.6924178674344362 | 0.0001
33 | 0.1250443458557129 | 0.2910602910602911 | 0.8133097762073027 | 0.6959916792345996 | 0.0001
34 | 0.12511762976646423 | 0.29417879417879417 | 0.8116187492060803 | 0.6891130310994343 | 0.0001
35 | 0.12488266825675964 | 0.2955647955647956 | 0.8124288545048274 | 0.6945448365895581 | 0.0001
36 | 0.1252983808517456 | 0.29417879417879417 | 0.8115410842141152 | 0.6971439978031583 | 0.0001
37 | 0.12479764968156815 | 0.29521829521829523 | 0.8116249469664828 | 0.6961006786941204 | 0.0001
38 | 0.12497606873512268 | 0.3004158004158004 | 0.8129930394431555 | 0.6991177533793484 | 0.0001
39 | 0.1252022236585617 | 0.29521829521829523 | 0.8141541282874172 | 0.6970545191351545 | 0.0001
40 | 0.12485132366418839 | 0.2955647955647956 | 0.816655585106383 | 0.7070171403235663 | 0.0001
41 | 0.12500154972076416 | 0.28967428967428965 | 0.8103573101656658 | 0.6961881266838973 | 0.0001
42 | 0.12350151687860489 | 0.3038808038808039 | 0.816535301022975 | 0.7064304960359926 | 0.0001
43 | 0.12367021292448044 | 0.2955647955647956 | 0.8150093808630394 | 0.7047254887418923 | 0.0001
44 | 0.12371324002742767 | 0.30076230076230076 | 0.8170209225905745 | 0.705396366545505 | 0.0001
45 | 0.12333343178033829 | 0.30145530145530147 | 0.8163231034048448 | 0.7058009223379548 | 0.0001
46 | 0.12297776341438293 | 0.30076230076230076 | 0.8158692722371967 | 0.6992655670184796 | 0.0001
47 | 0.12366960942745209 | 0.29902979902979904 | 0.8135392426486143 | 0.7026416067016249 | 0.0001
48 | 0.12326876819133759 | 0.30180180180180183 | 0.8169049621530698 | 0.7044430417074125 | 0.0001
49 | 0.12315386533737183 | 0.30214830214830213 | 0.8161126713333613 | 0.705026725915288 | 0.0001
50 | 0.12265044450759888 | 0.30145530145530147 | 0.8179686845851126 | 0.7085649491291086 | 0.0001
51 | 0.12310674786567688 | 0.30214830214830213 | 0.8190420609445996 | 0.710831288086539 | 0.0001
52 | 0.12280686944723129 | 0.30214830214830213 | 0.816390260370511 | 0.704117146056294 | 0.0001
53 | 0.1225290596485138 | 0.3038808038808039 | 0.8189015751312609 | 0.7080185810697228 | 0.0001
54 | 0.12376156449317932 | 0.30180180180180183 | 0.8162527837304089 | 0.7053875588266636 | 0.0001
55 | 0.12211860716342926 | 0.30284130284130284 | 0.818075117370892 | 0.7092508494713976 | 0.0001
56 | 0.12255053967237473 | 0.3049203049203049 | 0.818769689935334 | 0.7091508009521661 | 0.0001
57 | 0.12233822792768478 | 0.3052668052668053 | 0.8183564389510606 | 0.7056269081565454 | 0.0001
58 | 0.12230789661407471 | 0.30284130284130284 | 0.8179678964618875 | 0.7093876090831799 | 0.0001
59 | 0.12226579338312149 | 0.30734580734580735 | 0.8198051269184126 | 0.7102428483836337 | 0.0001
60 | 0.1236739531159401 | 0.29799029799029797 | 0.8173416232565955 | 0.7068409531794828 | 0.0001
61 | 0.12236195057630539 | 0.305959805959806 | 0.8201011747982775 | 0.7139384635953806 | 0.0001
62 | 0.12215279042720795 | 0.30284130284130284 | 0.8209334277030684 | 0.7188990083298508 | 1e-05
63 | 0.12084941565990448 | 0.3097713097713098 | 0.820752746564184 | 0.7190866276619315 | 1e-05
64 | 0.12093428522348404 | 0.3108108108108108 | 0.8218151540383014 | 0.7187730185556146 | 1e-05
65 | 0.12085793167352676 | 0.30803880803880807 | 0.8209837715435904 | 0.7186584702198188 | 1e-05
66 | 0.12076118588447571 | 0.3135828135828136 | 0.8215507887488523 | 0.7185770967712465 | 1e-05
67 | 0.1210499182343483 | 0.31115731115731116 | 0.8232429532417151 | 0.7239469969506999 | 1e-05
68 | 0.1208076998591423 | 0.3125433125433125 | 0.8211584808443447 | 0.720063006101889 | 1e-05
69 | 0.12105683237314224 | 0.31011781011781014 | 0.821014765549839 | 0.7197984794848579 | 1e-05
70 | 0.12111356854438782 | 0.31115731115731116 | 0.821309285237141 | 0.719699492247552 | 1e-05
71 | 0.12063230574131012 | 0.31115731115731116 | 0.8206033106461642 | 0.7163966165871272 | 1e-05
72 | 0.12075439840555191 | 0.3128898128898129 | 0.8206118081490495 | 0.7171818163524962 | 1e-05
73 | 0.12078637629747391 | 0.31323631323631324 | 0.8217462106977327 | 0.7214307826544399 | 1e-05
74 | 0.12086880952119827 | 0.3108108108108108 | 0.8200794388574326 | 0.715483654869702 | 1e-05
75 | 0.12054955214262009 | 0.3153153153153153 | 0.8207404925448148 | 0.7151281975948514 | 1e-05
76 | 0.12033110857009888 | 0.31566181566181567 | 0.8221261740503699 | 0.722403613960237 | 1e-05
77 | 0.12079885601997375 | 0.3135828135828136 | 0.8231996372480317 | 0.7234417953998725 | 1e-05
78 | 0.12099317461252213 | 0.3115038115038115 | 0.8230326613403982 | 0.7233107692667189 | 1e-05
79 | 0.12051720172166824 | 0.31011781011781014 | 0.8202369947054374 | 0.7172980311198172 | 1e-05
80 | 0.12073608487844467 | 0.31185031185031187 | 0.8231793006530544 | 0.7248558336823359 | 1e-05
81 | 0.12031927704811096 | 0.3128898128898129 | 0.822080253872813 | 0.7212996450160633 | 1e-05
82 | 0.1204884946346283 | 0.3142758142758143 | 0.8215302193202746 | 0.7178066335813648 | 1e-05
83 | 0.12136666476726532 | 0.31115731115731116 | 0.8179971218149497 | 0.7113142483409282 | 1.0000000000000002e-06
84 | 0.12041348963975906 | 0.3115038115038115 | 0.8234267187629895 | 0.7250649377579587 | 1.0000000000000002e-06
85 | 0.12035409361124039 | 0.31323631323631324 | 0.8229879338226147 | 0.7213085414821642 | 1.0000000000000002e-06
86 | 0.12250283360481262 | 0.3076923076923077 | 0.8196243388446962 | 0.7218120076279698 | 1.0000000000000002e-06
87 | 0.12075748294591904 | 0.3090783090783091 | 0.8203968852047224 | 0.7151954083158903 | 1.0000000000000002e-06
88 | 0.12086642533540726 | 0.30838530838530837 | 0.8215440749647566 | 0.7168335672232342 | 1.0000000000000002e-06
89 | 0.12105640023946762 | 0.3163548163548164 | 0.8244650323850127 | 0.733984551040518 | 1.0000000000000002e-07
90 | 0.12090421468019485 | 0.31185031185031187 | 0.8232248520710059 | 0.7245620055819162 | 1.0000000000000002e-07
91 | 0.12043782323598862 | 0.3115038115038115 | 0.8200938495056143 | 0.7163143946337084 | 1.0000000000000002e-07
---
# CO2 Emissions
The estimated CO2 emissions for training this model are documented below:
- **Emissions**: 1.029303722975925 grams of CO2
- **Source**: Code Carbon
- **Training Type**: fine-tuning
- **Geographical Location**: Brest, France
- **Hardware Used**: NVIDIA Tesla V100 PCIe 32 Go
---
# Framework Versions
- **Transformers**: 4.44.2
- **Pytorch**: 2.4.1+cu121
- **Datasets**: 3.0.0
- **Tokenizers**: 0.19.1
|