{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/danish-fungi-2020-not-just-another-image","title":"Danish Fungi 2020 -- Not Just Another Image Recognition Dataset","arxiv_id":"2103.10107","date":"2021-03-18","proceeding":null,"authors":["Lukáš Picek","Milan Šulc","Jiří Matas","Jacob Heilmann-Clausen","Thomas S. Jeppesen","Thomas Læssøe","Tobias Frøslev"],"abstract":"We introduce a novel fine-grained dataset and benchmark, the Danish Fungi 2020 (DF20). The dataset, constructed from observations submitted to the Atlas of Danish Fungi, is unique in its taxonomy-accurate class labels, small number of errors, highly unbalanced long-tailed class distribution, rich observation metadata, and well-defined class hierarchy. DF20 has zero overlap with ImageNet, allowing unbiased comparison of models fine-tuned from publicly available ImageNet checkpoints. The proposed evaluation protocol enables testing the ability to improve classification using metadata -- e.g. precise geographic location, habitat, and substrate, facilitates classifier calibration testing, and finally allows to study the impact of the device settings on the classification performance. Experiments using Convolutional Neural Networks (CNN) and the recent Vision Transformers (ViT) show that DF20 presents a challenging task. Interestingly, ViT achieves results superior to CNN baselines with 80.45% accuracy and 0.743 macro F1 score, reducing the CNN error by 9% and 12% respectively. A simple procedure for including metadata into the decision process improves the classification accuracy by more than 2.95 percentage points, reducing the error rate by 15%. The source code for all methods and experiments is available at https://sites.google.com/view/danish-fungi-dataset.","url_abs":"https://arxiv.org/abs/2103.10107v4","url_pdf":"https://arxiv.org/pdf/2103.10107v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"danish-fungi-2020-not-just-another-image","repo_url":"https://github.com/picekl/DanishFungiDataset","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classifier-calibration","task_name":"Classifier calibration"},{"task_slug":"fine-grained-image-classification","task_name":"Fine-Grained Image Classification"},{"task_slug":"fine-grained-image-recognition","task_name":"Fine-Grained Image Recognition"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"depthwise-convolution","method_name":"Depthwise Convolution"},{"method_slug":"depthwise-separable-convolution","method_name":"Depthwise Separable Convolution"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"efficientnet","method_name":"EfficientNet"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"grouped-convolution","method_name":"Grouped Convolution"},{"method_slug":"inception-a","method_name":"Inception-A"},{"method_slug":"inception-b","method_name":"Inception-B"},{"method_slug":"inception-c","method_name":"Inception-C"},{"method_slug":"inception-resnet-v2","method_name":"Inception-ResNet-v2"},{"method_slug":"inception-resnet-v2-reduction-b","method_name":"Inception-ResNet-v2 Reduction-B"},{"method_slug":"inception-resnet-v2-a","method_name":"Inception-ResNet-v2-A"},{"method_slug":"inception-resnet-v2-b","method_name":"Inception-ResNet-v2-B"},{"method_slug":"inception-resnet-v2-c","method_name":"Inception-ResNet-v2-C"},{"method_slug":"inception-v4","method_name":"Inception-v4"},{"method_slug":"inverted-residual-block","method_name":"Inverted Residual Block"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"pointwise-convolution","method_name":"Pointwise Convolution"},{"method_slug":"rmsprop","method_name":"RMSProp"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"reduction-a","method_name":"Reduction-A"},{"method_slug":"reduction-b","method_name":"Reduction-B"},{"method_slug":"resnext","method_name":"ResNeXt"},{"method_slug":"resnext-block","method_name":"ResNeXt Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"senet","method_name":"SENet"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"squeeze-and-excitation-block","method_name":"Squeeze-and-Excitation Block"},{"method_slug":"vision-transformer","method_name":"Vision Transformer"}],"datasets_introduced":[{"slug":"df20","name":"DF20","full_name":"Danish Fungi 2020"},{"slug":"df20-mini","name":"DF20 - Mini","full_name":"Danish Fungi 2020 - Mini"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ViT-Large/16 (384)","rank_in_archive_order":1,"of":19,"metrics":{"F1 - macro":"0.743","Top-1":"80.45","Top-3":"91.68"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ViT-Base/16 (384)","rank_in_archive_order":2,"of":19,"metrics":{"F1 - macro":"0.727","Top-1":"79.48","Top-3":"90.95"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"SE-ResNeXt-101-32x4d","rank_in_archive_order":3,"of":19,"metrics":{"Top-1":"77.13"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B5 (299)","rank_in_archive_order":4,"of":19,"metrics":{"F1 - macro":"0.678","Top-1":"76.1","Top-3":"88.85"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B3 (299)","rank_in_archive_order":5,"of":19,"metrics":{"F1 - macro":"0.673","Top-1":"75.69","Top-3":"88.72"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ViT-Large/16 (224)","rank_in_archive_order":6,"of":19,"metrics":{"F1 - macro":"0.675","Top-1":"75.29","Top-3":"88.34"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"SE-ResNeXt-101-32x4d (224)","rank_in_archive_order":7,"of":19,"metrics":{"F1 - macro":"0.66","Top-1":"74.26","Top-3":"87.78"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B1 (299)","rank_in_archive_order":8,"of":19,"metrics":{"F1 - macro":"0.654","Top-1":"74.08","Top-3":"87.68"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"Inception-ResNet-V2 (299)","rank_in_archive_order":9,"of":19,"metrics":{"F1 - macro":"0.651","Top-1":"74.01","Top-3":"87.49"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B0 (299)","rank_in_archive_order":10,"of":19,"metrics":{"Top-1":"73.65"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ResNet-50 (299)","rank_in_archive_order":11,"of":19,"metrics":{"Top-1":"73.49","Top-3":"87.13"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"Inception-V4 (299)","rank_in_archive_order":12,"of":19,"metrics":{"F1 - macro":"0.637","Top-1":"73","Top-3":"86.87"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B3 (224)","rank_in_archive_order":13,"of":19,"metrics":{"F1 - macro":"0.634","Top-1":"72.51","Top-3":"86.77"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"Inception-V3 (299)","rank_in_archive_order":14,"of":19,"metrics":{"Top-1":"72.1","Top-3":"86.58"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"EfficientNet-B0 (224)","rank_in_archive_order":15,"of":19,"metrics":{"F1 - macro":"0.613","Top-1":"70.33","Top-3":"85.19"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"MobileNet-V2 (299)","rank_in_archive_order":16,"of":19,"metrics":{"Top-1":"69.77","Top-3":"85.01"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ResNet-18","rank_in_archive_order":17,"of":19,"metrics":{"F1 - macro":"0.580","Top-1":"67.13","Top-3":"82.65"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"SE-ResNeXt-101-32x4d (299)","rank_in_archive_order":18,"of":19,"metrics":{"F1 - macro":"0.693","Top-3":"89.48"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20","task":"Image Classification","dataset":"DF20","model":"ResNet-34 (299)","rank_in_archive_order":19,"of":19,"metrics":{"F1 - macro":"0.60","Top-3":"84.76"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ViT-Large/16 (384)","rank_in_archive_order":1,"of":19,"metrics":{"F1 - macro":"0.669","Top-1":"75.85","Top-3":"89.95"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ViT-Base/16 (384)","rank_in_archive_order":2,"of":19,"metrics":{"F1 - macro":"0.639","Top-1":"74.23","Top-3":"89.12"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"SE-ResNeXt-101-32x4d","rank_in_archive_order":3,"of":19,"metrics":{"Top-1":"72.23"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ViT-Large/16 (224)","rank_in_archive_order":4,"of":19,"metrics":{"F1 - macro":"0.603","Top-1":"71.04","Top-3":"86.15"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B3 (299)","rank_in_archive_order":5,"of":19,"metrics":{"F1 - macro":"0.59","Top-1":"69.59","Top-3":"85.55"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"SE-ResNeXt-101-32x4d (224)","rank_in_archive_order":6,"of":19,"metrics":{"F1 - macro":"0.585","Top-1":"68.87","Top-3":"85.14"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B5 (299)","rank_in_archive_order":7,"of":19,"metrics":{"Top-1":"68.76","Top-3":"85"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ResNet-50 (299)","rank_in_archive_order":8,"of":19,"metrics":{"Top-1":"68.49","Top-3":"85.22"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B1 (299)","rank_in_archive_order":9,"of":19,"metrics":{"Top-1":"68.35","Top-3":"84.67"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B0 (299)","rank_in_archive_order":10,"of":19,"metrics":{"F1 - macro":"0.567","Top-1":"67.94","Top-3":"85.71"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"Inception-V4 (299)","rank_in_archive_order":11,"of":19,"metrics":{"Top-1":"67.45","Top-3":"82.78"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B3 (224)","rank_in_archive_order":12,"of":19,"metrics":{"F1 - macro":"0.55","Top-1":"67.39","Top-3":"83.74"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"Inception-V3 (299)","rank_in_archive_order":13,"of":19,"metrics":{"F1 - macro":"0.535","Top-1":"65.91","Top-3":"82.97"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"EfficientNet-B0 (224)","rank_in_archive_order":14,"of":19,"metrics":{"F1 - macro":"0.531","Top-1":"65.66","Top-3":"83.65"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"MobileNet-V2 (299)","rank_in_archive_order":15,"of":19,"metrics":{"Top-1":"65.58"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"Inception-ResNet-V2 (299)","rank_in_archive_order":16,"of":19,"metrics":{"Top-1":"64.67","Top-3":"81.42"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ResNet-18","rank_in_archive_order":17,"of":19,"metrics":{"F1 - macro":"0.514","Top-1":"62.91","Top-3":"81.65"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"SE-ResNeXt-101-32x4d (299)","rank_in_archive_order":18,"of":19,"metrics":{"F1 - macro":"0.62","Top-3":"87.28"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-df20-mini","task":"Image Classification","dataset":"DF20 - Mini","model":"ResNet-34 (299)","rank_in_archive_order":19,"of":19,"metrics":{"F1 - macro":"0.559","Top-3":"83.52"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2103.10107","atlas_url":"https://app.syntology.ai/?focus=2103.10107","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}