File size: 15,304 Bytes
4c96f1c
78598cf
4c96f1c
 
 
78598cf
4c96f1c
 
78598cf
4c96f1c
78598cf
 
 
 
 
4c96f1c
78598cf
 
 
 
 
 
4c96f1c
 
 
 
78598cf
4c96f1c
78598cf
4c96f1c
78598cf
4c96f1c
 
 
 
 
 
78598cf
4c96f1c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78598cf
4c96f1c
78598cf
4c96f1c
78598cf
4c96f1c
78598cf
 
4c96f1c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
232
233
234
235
236
237
238
239
240
241
242
243
244
245

---
language:
- eng
license: wtfpl
tags:
- multilabel-image-classification
- multilabel
- generated_from_trainer
base_model: facebook/dinov2-large
model-index:
- name: DinoVdeauTest-large-2024_09_24-batch-size32_freeze
  results: []
---

DinoVdeauTest 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: 0.1204
- F1 Micro: 0.8214
- F1 Macro: 0.7191
- Accuracy: 0.3135

---

# Model description
DinoVdeauTest is a model built on top of facebook/dinov2-large 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**: 106.0
- **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 | 0.17755262553691864 | 0.2169092169092169 | 0.7485599799649386 | 0.5520222156367808 | 0.001
2 | 0.15282750129699707 | 0.24047124047124047 | 0.7693301744561706 | 0.5698350844395493 | 0.001
3 | 0.14834792912006378 | 0.2494802494802495 | 0.7773639211038686 | 0.6216984689952817 | 0.001
4 | 0.1467081755399704 | 0.2553707553707554 | 0.777158535539741 | 0.6272242063327677 | 0.001
5 | 0.1453460305929184 | 0.24636174636174638 | 0.7772219080349763 | 0.6281234234815563 | 0.001
6 | 0.14370940625667572 | 0.262993762993763 | 0.7809704878364948 | 0.6191104313765434 | 0.001
7 | 0.14275749027729034 | 0.2598752598752599 | 0.7811054928907905 | 0.6172423187752968 | 0.001
8 | 0.1432911604642868 | 0.25571725571725573 | 0.7799277605779154 | 0.6252572102986063 | 0.001
9 | 0.14202255010604858 | 0.25190575190575193 | 0.7892645815722739 | 0.6529741984810739 | 0.001
10 | 0.24619852006435394 | 0.2553707553707554 | 0.7441173839133935 | 0.6021759252294137 | 0.001
11 | 0.14001792669296265 | 0.2702702702702703 | 0.7882727742379886 | 0.6483450509947135 | 0.001
12 | 0.13999028503894806 | 0.2598752598752599 | 0.7905884326636748 | 0.6590290067827226 | 0.001
13 | 0.13937941193580627 | 0.2643797643797644 | 0.7875647668393783 | 0.6466696619852873 | 0.001
14 | 0.13986562192440033 | 0.26403326403326405 | 0.7879462374796679 | 0.6469191780902835 | 0.001
15 | 0.13914281129837036 | 0.2512127512127512 | 0.7881215116526777 | 0.6413408300553888 | 0.001
16 | 0.14137022197246552 | 0.26334026334026334 | 0.7906518010291596 | 0.6347762053394217 | 0.001
17 | 0.14031356573104858 | 0.2616077616077616 | 0.7903470350404312 | 0.6445091939759507 | 0.001
18 | 0.14039942622184753 | 0.26022176022176025 | 0.792909822326517 | 0.6463740590949281 | 0.001
19 | 0.14035724103450775 | 0.25848925848925847 | 0.7935513900055008 | 0.6526097604635809 | 0.001
20 | 0.1390993297100067 | 0.255024255024255 | 0.7900462566386842 | 0.6491113591903472 | 0.001
21 | 0.13827534019947052 | 0.25814275814275817 | 0.7916862276471341 | 0.6507882842254223 | 0.001
22 | 0.13889800012111664 | 0.26853776853776856 | 0.7935178133197935 | 0.6489112697391156 | 0.001
23 | 0.1385144591331482 | 0.2560637560637561 | 0.7842915050342152 | 0.6549504216490796 | 0.001
24 | 0.13674499094486237 | 0.2695772695772696 | 0.7921174652241113 | 0.6507209930045859 | 0.001
25 | 0.13791973888874054 | 0.2751212751212751 | 0.7893792922444337 | 0.6408412361180978 | 0.001
26 | 0.1375003606081009 | 0.27096327096327094 | 0.7942844956280657 | 0.6468889221342444 | 0.001
27 | 0.13920167088508606 | 0.2643797643797644 | 0.7921208479958802 | 0.6516252219978993 | 0.001
28 | 0.13846370577812195 | 0.2636867636867637 | 0.7917504599717599 | 0.6424574000169906 | 0.001
29 | 0.14017289876937866 | 0.2626472626472626 | 0.7883173722754258 | 0.6496760555529415 | 0.001
30 | 0.13773277401924133 | 0.2668052668052668 | 0.788684508761224 | 0.6553418932728605 | 0.001
31 | 0.1312580555677414 | 0.27893277893277896 | 0.8017825006427286 | 0.6718196714045327 | 0.0001
32 | 0.1318267583847046 | 0.2806652806652807 | 0.8060657118786858 | 0.6772473958552189 | 0.0001
33 | 0.1309264600276947 | 0.27754677754677753 | 0.8050148746281343 | 0.6791810095569879 | 0.0001
34 | 0.1296369731426239 | 0.28205128205128205 | 0.804873906767428 | 0.6774879025186358 | 0.0001
35 | 0.12816031277179718 | 0.28932778932778935 | 0.8084945013272659 | 0.6864816840760736 | 0.0001
36 | 0.12886497378349304 | 0.2830907830907831 | 0.8055151151836945 | 0.6827519815644174 | 0.0001
37 | 0.12768980860710144 | 0.2830907830907831 | 0.8054549328787556 | 0.677476009100308 | 0.0001
38 | 0.1274958997964859 | 0.288981288981289 | 0.8084112149532711 | 0.6882635597933275 | 0.0001
39 | 0.12658803164958954 | 0.28794178794178793 | 0.8099270562589084 | 0.6853651532137502 | 0.0001
40 | 0.12820227444171906 | 0.28863478863478864 | 0.8117343111925994 | 0.698146081234717 | 0.0001
41 | 0.12668322026729584 | 0.2882882882882883 | 0.8081688455951466 | 0.6850722645729148 | 0.0001
42 | 0.12620903551578522 | 0.2907137907137907 | 0.8112331081081081 | 0.6941693850580287 | 0.0001
43 | 0.12590545415878296 | 0.2910602910602911 | 0.8107360033550011 | 0.69077693559172 | 0.0001
44 | 0.12642185389995575 | 0.29313929313929316 | 0.8114530416894076 | 0.6925170228864211 | 0.0001
45 | 0.1257808804512024 | 0.2966042966042966 | 0.8110017663386323 | 0.6974852591826903 | 0.0001
46 | 0.12542255222797394 | 0.2986832986832987 | 0.8109046268467172 | 0.694112624204446 | 0.0001
47 | 0.1256514936685562 | 0.2920997920997921 | 0.8098117995347853 | 0.6937033329875891 | 0.0001
48 | 0.1253652125597 | 0.29140679140679143 | 0.8106740862526874 | 0.6904792900172609 | 0.0001
49 | 0.1252448409795761 | 0.2945252945252945 | 0.8136526746223817 | 0.6973960572840976 | 0.0001
50 | 0.12482810020446777 | 0.2983367983367983 | 0.814972104255142 | 0.7025806374316313 | 0.0001
51 | 0.12458823621273041 | 0.2959112959112959 | 0.815773054365459 | 0.7066639116616246 | 0.0001
52 | 0.12461517751216888 | 0.29521829521829523 | 0.8120651620566774 | 0.7008785276837475 | 0.0001
53 | 0.12417320907115936 | 0.29902979902979904 | 0.8142762173840784 | 0.697381421827107 | 0.0001
54 | 0.12411519885063171 | 0.2966042966042966 | 0.8135221716405328 | 0.7000554609843794 | 0.0001
55 | 0.12417341023683548 | 0.29521829521829523 | 0.8130868668154448 | 0.6996806217177095 | 0.0001
56 | 0.1234845370054245 | 0.30214830214830213 | 0.8178580303155802 | 0.7064033143378504 | 0.0001
57 | 0.12350637465715408 | 0.2993762993762994 | 0.8150138550675959 | 0.6963082116005005 | 0.0001
58 | 0.12311282008886337 | 0.2983367983367983 | 0.814454598809424 | 0.7011763866264856 | 0.0001
59 | 0.12340909987688065 | 0.30006930006930005 | 0.8153207895067678 | 0.7021882450681799 | 0.0001
60 | 0.12386388331651688 | 0.2972972972972973 | 0.8121788610981358 | 0.6978379224258093 | 0.0001
61 | 0.12355918437242508 | 0.30145530145530147 | 0.8157532819337362 | 0.7114090358686158 | 0.0001
62 | 0.12266429513692856 | 0.3031878031878032 | 0.8168304358780549 | 0.7119609165997448 | 0.0001
63 | 0.12305888533592224 | 0.2948717948717949 | 0.8136965505608501 | 0.7077249537938423 | 0.0001
64 | 0.12276986986398697 | 0.30561330561330563 | 0.8172187094342783 | 0.708440477929098 | 0.0001
65 | 0.12323789298534393 | 0.3076923076923077 | 0.8182651445712857 | 0.7103116175502073 | 0.0001
66 | 0.12264254689216614 | 0.30353430353430355 | 0.8178908425822006 | 0.706549053618703 | 0.0001
67 | 0.12282951176166534 | 0.3052668052668053 | 0.8184709429598117 | 0.7104724530407897 | 0.0001
68 | 0.12277437746524811 | 0.3042273042273042 | 0.818125541661508 | 0.7127667888432785 | 0.0001
69 | 0.12282923609018326 | 0.3052668052668053 | 0.8137196924896511 | 0.703803449192515 | 0.0001
70 | 0.1231524795293808 | 0.30180180180180183 | 0.8154847434705434 | 0.7079779831387449 | 0.0001
71 | 0.12310803681612015 | 0.29902979902979904 | 0.8155765340525961 | 0.7111485120209401 | 0.0001
72 | 0.12233748286962509 | 0.30076230076230076 | 0.8161826422695988 | 0.7149554589558469 | 0.0001
73 | 0.12232980877161026 | 0.3049203049203049 | 0.8173891171292027 | 0.7041677932151664 | 0.0001
74 | 0.12370481342077255 | 0.29625779625779625 | 0.8124655033329088 | 0.7009378784168091 | 0.0001
75 | 0.12251079827547073 | 0.30457380457380456 | 0.8151614807319335 | 0.7045397920320168 | 0.0001
76 | 0.12471619248390198 | 0.30076230076230076 | 0.8159596808063839 | 0.7099467053038022 | 0.0001
77 | 0.1225149855017662 | 0.29902979902979904 | 0.8179482930062977 | 0.7139220834888879 | 0.0001
78 | 0.12218895554542542 | 0.30457380457380456 | 0.8188081009296149 | 0.7060530834173799 | 0.0001
79 | 0.12460680305957794 | 0.30180180180180183 | 0.8152018201735907 | 0.7100736740896743 | 0.0001
80 | 0.12210072576999664 | 0.3038808038808039 | 0.8179503235232728 | 0.710254660316209 | 0.0001
81 | 0.12116113305091858 | 0.30180180180180183 | 0.8184843191082273 | 0.7156736962695093 | 0.0001
82 | 0.12159755080938339 | 0.30803880803880807 | 0.8151834668916069 | 0.7088973142030256 | 0.0001
83 | 0.12142453342676163 | 0.30803880803880807 | 0.8164913756836348 | 0.7090314532601811 | 0.0001
84 | 0.12157817929983139 | 0.306999306999307 | 0.8168789808917197 | 0.7100088286352488 | 0.0001
85 | 0.12155645340681076 | 0.3052668052668053 | 0.8188048474717927 | 0.7109093140233731 | 0.0001
86 | 0.12185127288103104 | 0.306999306999307 | 0.8190986316274009 | 0.7176309861117636 | 0.0001
87 | 0.12196006625890732 | 0.3063063063063063 | 0.8176601181250785 | 0.7080074645247395 | 0.0001
88 | 0.1209772601723671 | 0.3049203049203049 | 0.8210709982967056 | 0.7158455151348946 | 1e-05
89 | 0.12103869765996933 | 0.30734580734580735 | 0.82409381663113 | 0.7312085502704033 | 1e-05
90 | 0.12060839682817459 | 0.306999306999307 | 0.823466447097571 | 0.7232165780756572 | 1e-05
91 | 0.12036494165658951 | 0.3087318087318087 | 0.8190853196327803 | 0.714800773362679 | 1e-05
92 | 0.12028194963932037 | 0.3087318087318087 | 0.8194187804468336 | 0.7131694962358415 | 1e-05
93 | 0.12036142498254776 | 0.30838530838530837 | 0.8214921910601102 | 0.7183857334665726 | 1e-05
94 | 0.12012535333633423 | 0.30665280665280664 | 0.8207247828991316 | 0.7195089038891936 | 1e-05
95 | 0.12010551244020462 | 0.30838530838530837 | 0.8197421299397187 | 0.7157953029268052 | 1e-05
96 | 0.1198035180568695 | 0.3052668052668053 | 0.8217099503939306 | 0.7193329750108997 | 1e-05
97 | 0.12007978558540344 | 0.3063063063063063 | 0.8204592028773368 | 0.7210913273742954 | 1e-05
98 | 0.12013950198888779 | 0.30803880803880807 | 0.8226615276223631 | 0.7250050738866957 | 1e-05
99 | 0.12003140896558762 | 0.30734580734580735 | 0.8206050968740846 | 0.7226446473273916 | 1e-05
100 | 0.1200241968035698 | 0.30803880803880807 | 0.8216021451315569 | 0.7191058080317732 | 1e-05
101 | 0.11989927291870117 | 0.3097713097713098 | 0.8228338260398884 | 0.724185821594473 | 1e-05
102 | 0.12043397128582001 | 0.30734580734580735 | 0.8192071374463429 | 0.7216480982679994 | 1e-05
103 | 0.1199527159333229 | 0.30665280665280664 | 0.8221990402670561 | 0.7231653777895258 | 1.0000000000000002e-06
104 | 0.12043838202953339 | 0.31011781011781014 | 0.8217751676454663 | 0.7234886652066129 | 1.0000000000000002e-06
105 | 0.11985885351896286 | 0.3076923076923077 | 0.8219212232009308 | 0.7205886261735153 | 1.0000000000000002e-06
106 | 0.11984959244728088 | 0.31046431046431044 | 0.8228195739014648 | 0.7270103073654017 | 1.0000000000000002e-06


---

# CO2 Emissions

The estimated CO2 emissions for training this model are documented below:

- **Emissions**: 1.1530867349091423 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