|
|
|
--- |
|
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 |
|
|
|
|
