# Hierarchical Image Classification on iNaturalist 2021 Source: [Hierarchical Image Classification on iNaturalist 2021]() Collection index: [llms.txt]() Data updated: 2026-06-23T09:40:04.286Z Evaluation of Hierarchical Image Classification performance on the iNaturalist 2021 dataset, utilizing metrics across coarse and fine classification levels. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Hierarchical Image Classification | | Dataset (including any stated split) | iNaturalist 2021 | | Metric | Accuracy (fine) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy (fine)](): Higher is better; unit: PERCENT. - [Accuracy (coarse)](): Higher is better; unit: PERCENT. - [Log Consistency Score](): Higher is better; unit: SCALAR. - [Precision (coarse)](): Higher is better; unit: PERCENT. - [Recall (coarse)](): Higher is better; unit: PERCENT. - [F1 Score (coarse)](): Higher is better; unit: PERCENT. - [ECE (fine)](): Lower is better; unit: SCALAR. - [ECE (coarse)](): Lower is better; unit: SCALAR. - [Entropy (fine)](): Lower is better; unit: SCALAR. - [Entropy (coarse)](): Lower is better; unit: SCALAR. - [Accuracy (Kingdom)](): Higher is better; unit: PERCENT. - [Accuracy (Phylum)](): Higher is better; unit: PERCENT. - [Accuracy (Class)](): Higher is better; unit: PERCENT. - [Accuracy (Order)](): Higher is better; unit: PERCENT. - [Accuracy (Family)](): Higher is better; unit: PERCENT. - [Accuracy (Genus)](): Higher is better; unit: PERCENT. - [Accuracy (Overall Avg)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (fine) | Result date | Report paper | | --- | --- | --- | --- | | NeSy Epistemic | 85.12 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 84.57 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 84.42 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 83.65 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 83.19 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 82.46 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 82.29 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 82.17 | Not recorded | 6a0bee86dbc836b87d82c351 | | NeSy Epistemic | 81.97 | Not recorded | 6a0bee86dbc836b87d82c351 | | Epistemic (RS-NN) | 79.55 | Not recorded | 6a0bee86dbc836b87d82c351 | | Base | 79.04 | Not recorded | 6a0bee86dbc836b87d82c351 | | Level-Restricted Contrastive Learning | 63 | Not recorded | 6a3a1324206542251d7c70bd | | Level-Restricted Contrastive Learning | 51.35 | Not recorded | 6a3a1324206542251d7c70bd | | BioCLIP | 50.79 | Not recorded | 6a3a1324206542251d7c70bd | | RCME | 50.5 | Not recorded | 6a3a1324206542251d7c70bd | | OpenCLIP | 2.09 | Not recorded | 6a3a1324206542251d7c70bd | ### NeSy Epistemic Accuracy (fine): 85.12. Result ID: 6a0c0cf7fbc373592c019af3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Lukasiewicz","Membership function":"Gaussian","tau\_f":"0.6","tau\_c":"0.5","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 84.57. Result ID: 6a0c0cf7fbc373592c019aee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Godel","Membership function":"Trapezoidal","tau\_f":"0.4","tau\_c":"0.5","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 84.42. Result ID: 6a0c0cf7fbc373592c019af0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Product","Membership function":"Triangular","tau\_f":"0.4","tau\_c":"0.5","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 83.65. Result ID: 6a0c0cf7fbc373592c019af1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Product","Membership function":"Trapezoidal","tau\_f":"0.4","tau\_c":"0.5","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 83.19. Result ID: 6a0c0cf7fbc373592c019af4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Lukasiewicz","Membership function":"Trapezoidal","tau\_f":"0.4","tau\_c":"0.5","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 82.46. Result ID: 6a0c0cf7fbc373592c019af5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Lukasiewicz","Membership function":"Triangular","tau\_f":"0.4","tau\_c":"0.4","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 82.29. Result ID: 6a0c0cf7fbc373592c019af2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Product","Membership function":"Gaussian","tau\_f":"0.4","tau\_c":"0.4","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 82.17. Result ID: 6a0c0cf7fbc373592c019aed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Godel","Membership function":"Gaussian","tau\_f":"0.4","tau\_c":"0.4","Warm up":"true"}. This record does not assert independent reproduction. ### NeSy Epistemic Accuracy (fine): 81.97. Result ID: 6a0c0cf7fbc373592c019aef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"t-norm family":"Godel","Membership function":"Triangular","tau\_f":"0.4","tau\_c":"0.4","Warm up":"true"}. This record does not assert independent reproduction. ### Epistemic (RS-NN) Accuracy (fine): 79.55. Result ID: 6a0c0cf7fbc373592c019aec. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Base Accuracy (fine): 79.04. Result ID: 6a0c0cf7fbc373592c019aeb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Level-Restricted Contrastive Learning Accuracy (fine): 63. Result ID: 6a3a5464fbc373592c35fbda. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Space":"Euclidean"}. This record does not assert independent reproduction. ### Level-Restricted Contrastive Learning Accuracy (fine): 51.35. Result ID: 6a3a5464fbc373592c35fbdb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Space":"Hyperbolic"}. This record does not assert independent reproduction. ### BioCLIP Accuracy (fine): 50.79. Result ID: 6a3a5464fbc373592c35fbd8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Space":"Euclidean"}. This record does not assert independent reproduction. ### RCME Accuracy (fine): 50.5. Result ID: 6a3a5464fbc373592c35fbd9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Space":"Euclidean"}. This record does not assert independent reproduction. ### OpenCLIP Accuracy (fine): 2.09. Result ID: 6a3a5464fbc373592c35fbd7. [Reporting paper]() Recorded evaluation configuration: {"Space":"Euclidean"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.