# Novel Class Discovery on SemanticPOSS (Split 3) Source: [Novel Class Discovery on SemanticPOSS (Split 3)]() Collection index: [llms.txt]() Data updated: 2026-06-08T07:46:10.922Z Evaluation of Novel Class Discovery performance on Split 3 of the SemanticPOSS dataset, reporting metrics for known, novel, and overall classes, alongside class-specific scores for various object categories. 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 | Novel Class Discovery | | Dataset (including any stated split) | SemanticPOSS (Split 3) | | Metric | Bike Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 5 of 5 | | 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 - [Bike Accuracy](): Higher is better; unit: PERCENT. - [Building Accuracy](): Higher is better; unit: PERCENT. - [Car Accuracy](): Higher is better; unit: PERCENT. - [Cone Accuracy](): Higher is better; unit: PERCENT. - [Fence Accuracy](): Higher is better; unit: PERCENT. - [Ground Accuracy](): Higher is better; unit: PERCENT. - [Person Accuracy](): Higher is better; unit: PERCENT. - [Plants Accuracy](): Higher is better; unit: PERCENT. - [Pole Accuracy](): Higher is better; unit: PERCENT. - [Rider Accuracy](): Higher is better; unit: PERCENT. - [Traffic Sign Accuracy](): Higher is better; unit: PERCENT. - [Trash Can Accuracy](): Higher is better; unit: PERCENT. - [Trunk Accuracy](): Higher is better; unit: PERCENT. - [Novel Class Accuracy](): Higher is better; unit: PERCENT. - [Known Class Accuracy](): Higher is better; unit: PERCENT. - [Overall Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Bike Accuracy | Result date | Report paper | | --- | --- | --- | --- | | DASL | 45.5 | Not recorded | 6a264ca690966ade919beae2 | | HyperNCD | 44.2 | Not recorded | 6a264ca690966ade919beae2 | | EUMS | 41.2 | Not recorded | 6a264ca690966ade919beae2 | | SNOPS | 39.4 | Not recorded | 6a264ca690966ade919beae2 | | NOPS | 38.6 | Not recorded | 6a264ca690966ade919beae2 | ### DASL Bike Accuracy: 45.5. Result ID: 6a26733ffbc373592c22f5a7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MinkowskiUNet-34C","Learning Protocol":"Novel Class Discovery"}. This record does not assert independent reproduction. ### HyperNCD Bike Accuracy: 44.2. Result ID: 6a26733ffbc373592c22f5a8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MinkowskiUNet-34C","Learning Protocol":"Novel Class Discovery"}. This record does not assert independent reproduction. ### EUMS Bike Accuracy: 41.2. Result ID: 6a26733ffbc373592c22f5a4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MinkowskiUNet-34C","Learning Protocol":"Novel Class Discovery"}. This record does not assert independent reproduction. ### SNOPS Bike Accuracy: 39.4. Result ID: 6a26733ffbc373592c22f5a6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MinkowskiUNet-34C","Learning Protocol":"Novel Class Discovery"}. This record does not assert independent reproduction. ### NOPS Bike Accuracy: 38.6. Result ID: 6a26733ffbc373592c22f5a5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"MinkowskiUNet-34C","Learning Protocol":"Novel Class Discovery"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.