--- language: - eng license: wtfpl tags: - multilabel-image-classification - multilabel - generated_from_trainer base_model: facebook/dinov2-base model-index: - name: dinov2-base-2024_09_09-batch-size32_epochs150_freeze results: [] --- DinoVd'eau is a fine-tuned version of [facebook/dinov2-base](https://huggingface.co/facebook/dinov2-base). It achieves the following results on the test set: - Loss: 0.1321 - F1 Micro: 0.8069 - F1 Macro: 0.7121 - Roc Auc: 0.8742 - Accuracy: 0.2869 --- # 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 | 0.16006726026535034 | 0.23284823284823286 | 0.7633800438966739 | 0.6250897780499145 | 0.001 2 | 0.150440976023674 | 0.24982674982674982 | 0.7780064686856808 | 0.646165211379598 | 0.001 3 | 0.14829224348068237 | 0.2564102564102564 | 0.7816936696175046 | 0.6644318154557648 | 0.001 4 | 0.14641565084457397 | 0.2553707553707554 | 0.7862639635912287 | 0.680888104485521 | 0.001 5 | 0.14226503670215607 | 0.2681912681912682 | 0.7891243298442687 | 0.6919100708566497 | 0.001 6 | 0.1439608633518219 | 0.26507276507276506 | 0.7901946045268521 | 0.6987715680115144 | 0.001 7 | 0.1425073742866516 | 0.2681912681912682 | 0.7937821236053655 | 0.6849790066180481 | 0.001 8 | 0.14294348657131195 | 0.2636867636867637 | 0.793083667950504 | 0.6880365824342907 | 0.001 9 | 0.14630228281021118 | 0.25571725571725573 | 0.7926595005517636 | 0.6884565577441364 | 0.001 10 | 0.13922064006328583 | 0.27442827442827444 | 0.8009224940284985 | 0.7049759390767861 | 0.001 11 | 0.14429208636283875 | 0.26992376992376993 | 0.785345272946444 | 0.6892328865834217 | 0.001 12 | 0.14520499110221863 | 0.2713097713097713 | 0.7888341543513957 | 0.6976448599197044 | 0.001 13 | 0.13695523142814636 | 0.2765072765072765 | 0.8007200870802982 | 0.7032121010324246 | 0.001 14 | 0.14012356102466583 | 0.273042273042273 | 0.7983576642335767 | 0.6875097222118577 | 0.001 15 | 0.13785772025585175 | 0.2817047817047817 | 0.8048810652595126 | 0.7001361694791496 | 0.001 16 | 0.1429404616355896 | 0.2681912681912682 | 0.7968854097268487 | 0.7063273106998997 | 0.001 17 | 0.1451471894979477 | 0.26126126126126126 | 0.7956287718153646 | 0.6860743816280108 | 0.001 18 | 0.141770601272583 | 0.2713097713097713 | 0.7906203368151778 | 0.6849355289660601 | 0.001 19 | 0.14384245872497559 | 0.2654192654192654 | 0.7899699957136733 | 0.6794374521554336 | 0.001 20 | 0.13193023204803467 | 0.28655578655578656 | 0.8068363147728227 | 0.7201978132992005 | 0.0001 21 | 0.13121400773525238 | 0.2875952875952876 | 0.8080536912751679 | 0.7236910659256566 | 0.0001 22 | 0.1310088187456131 | 0.2934857934857935 | 0.810120343368793 | 0.7222147145142929 | 0.0001 23 | 0.1304517388343811 | 0.2934857934857935 | 0.8120394137616957 | 0.7226400439644629 | 0.0001 24 | 0.13093852996826172 | 0.29521829521829523 | 0.8096162584162916 | 0.7237916982943077 | 0.0001 25 | 0.13081994652748108 | 0.2948717948717949 | 0.8093388464269307 | 0.7170657451815683 | 0.0001 26 | 0.13007444143295288 | 0.2910602910602911 | 0.8099862459884133 | 0.7200172245050901 | 0.0001 27 | 0.13034380972385406 | 0.29244629244629244 | 0.8082065853250877 | 0.7207907434740295 | 0.0001 28 | 0.13018907606601715 | 0.29695079695079696 | 0.810349848163401 | 0.7217805682073449 | 0.0001 29 | 0.13019531965255737 | 0.29625779625779625 | 0.8104190823256585 | 0.723719101965087 | 0.0001 30 | 0.13030356168746948 | 0.2955647955647956 | 0.8096606287736832 | 0.718144679800513 | 0.0001 31 | 0.1301266849040985 | 0.2959112959112959 | 0.8092418049879057 | 0.7189603352791966 | 0.0001 32 | 0.1301257312297821 | 0.2927927927927928 | 0.8097980303789017 | 0.7210148516296496 | 0.0001 33 | 0.12959885597229004 | 0.29625779625779625 | 0.8099594769603543 | 0.7204264964359948 | 1e-05 34 | 0.12959957122802734 | 0.2955647955647956 | 0.8100854344655136 | 0.722168676552786 | 1e-05 35 | 0.12954092025756836 | 0.2955647955647956 | 0.8108894430590192 | 0.7220033007887567 | 1e-05 36 | 0.12953610718250275 | 0.29313929313929316 | 0.8104569713142095 | 0.7211650841899886 | 1e-05 37 | 0.1295497566461563 | 0.29625779625779625 | 0.8118778893007372 | 0.7239071903959954 | 1e-05 38 | 0.12949061393737793 | 0.2959112959112959 | 0.8104318798247445 | 0.7212977755433345 | 1e-05 39 | 0.12945865094661713 | 0.2966042966042966 | 0.8106218263547823 | 0.7221707642640621 | 1e-05 40 | 0.12946291267871857 | 0.2955647955647956 | 0.8113418729013804 | 0.7232749192333074 | 1e-05 41 | 0.1294611394405365 | 0.2945252945252945 | 0.8100071001962995 | 0.722313917509489 | 1e-05 42 | 0.12951640784740448 | 0.2972972972972973 | 0.8111398315684148 | 0.7219276596712088 | 1e-05 43 | 0.12940654158592224 | 0.29313929313929316 | 0.8097862391449566 | 0.7212066160587719 | 1e-05 44 | 0.12948854267597198 | 0.29695079695079696 | 0.8108311081441923 | 0.7211905265653523 | 1e-05 45 | 0.12943118810653687 | 0.2945252945252945 | 0.8103943697164036 | 0.7217673828508766 | 1e-05 46 | 0.12941767275333405 | 0.29764379764379767 | 0.8113435070065285 | 0.7232663108413819 | 1e-05 47 | 0.12936843931674957 | 0.2945252945252945 | 0.8107185952648442 | 0.7229077354567445 | 1e-05 48 | 0.12944123148918152 | 0.2955647955647956 | 0.8102512730611904 | 0.7208766406208041 | 1e-05 49 | 0.12932655215263367 | 0.2959112959112959 | 0.8111032502392942 | 0.7215165769975259 | 1e-05 50 | 0.1294257938861847 | 0.2966042966042966 | 0.8106959890041235 | 0.7210862927892402 | 1e-05 51 | 0.12937645614147186 | 0.29244629244629244 | 0.8098573930447837 | 0.7224236625273444 | 1e-05 52 | 0.12941104173660278 | 0.2972972972972973 | 0.8110019973368842 | 0.7223932851056244 | 1e-05 53 | 0.12947481870651245 | 0.29799029799029797 | 0.8110783049860689 | 0.7225026360610024 | 1e-05 54 | 0.12942463159561157 | 0.29625779625779625 | 0.8104531646623112 | 0.7221711249170111 | 1e-05 55 | 0.12934881448745728 | 0.2955647955647956 | 0.8107163657542226 | 0.7231230527181782 | 1e-05 56 | 0.12935101985931396 | 0.2959112959112959 | 0.810738813735692 | 0.7226955143721722 | 1.0000000000000002e-06 57 | 0.12934598326683044 | 0.2955647955647956 | 0.8110560712650376 | 0.7230703100391168 | 1.0000000000000002e-06 58 | 0.1293543428182602 | 0.2966042966042966 | 0.8112406328059951 | 0.7230017316798248 | 1.0000000000000002e-06 59 | 0.12936049699783325 | 0.2966042966042966 | 0.8110088687179914 | 0.7227156089091311 | 1.0000000000000002e-06 --- # CO2 Emissions The estimated CO2 emissions for training this model are documented below: - **Emissions**: 0.7291228651023076 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**: 4.41.1 - **Pytorch**: 2.3.0+cu121 - **Datasets**: 2.19.1 - **Tokenizers**: 0.19.1