# 3D Object Detection on SUN RGB-D (AP@0.15) Source: [3D Object Detection on SUN RGB-D (AP@0.15)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:23:21.861Z Evaluation of 3D object detection performance on the SUN RGB-D dataset, reporting class-specific Average Precision (AP) calculated specifically at an Intersection over Union (IoU) threshold of 0.15. 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 | 3D Object Detection | | Dataset (including any stated split) | SUN RGB-D | | Metric | AP@0.15 (Overall) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 7 of 7 | | 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 - [AP@0.15 (Overall)](): Higher is better; unit: PERCENT. - [AP@0.15 (cabinet)](): Higher is better; unit: PERCENT. - [AP@0.15 (bed)](): Higher is better; unit: PERCENT. - [AP@0.15 (chair)](): Higher is better; unit: PERCENT. - [AP@0.15 (sofa)](): Higher is better; unit: PERCENT. - [AP@0.15 (table)](): Higher is better; unit: PERCENT. - [AP@0.15 (door)](): Higher is better; unit: PERCENT. - [AP@0.15 (window)](): Higher is better; unit: PERCENT. - [AP@0.15 (book shelf)](): Higher is better; unit: PERCENT. - [AP@0.15 (picture)](): Higher is better; unit: PERCENT. - [AP@0.15 (counter)](): Higher is better; unit: PERCENT. - [AP@0.15 (blinds)](): Higher is better; unit: PERCENT. - [AP@0.15 (desk)](): Higher is better; unit: PERCENT. - [AP@0.15 (shelves)](): Higher is better; unit: PERCENT. - [AP@0.15 (curtain)](): Higher is better; unit: PERCENT. - [AP@0.15 (dresser)](): Higher is better; unit: PERCENT. - [AP@0.15 (pillow)](): Higher is better; unit: PERCENT. - [AP@0.15 (mirror)](): Higher is better; unit: PERCENT. - [AP@0.15 (floor mat)](): Higher is better; unit: PERCENT. - [AP@0.15 (clothes)](): Higher is better; unit: PERCENT. - [AP@0.15 (books)](): Higher is better; unit: PERCENT. - [AP@0.15 (fridge)](): Higher is better; unit: PERCENT. - [AP@0.15 (tv)](): Higher is better; unit: PERCENT. - [AP@0.15 (paper)](): Higher is better; unit: PERCENT. - [AP@0.15 (towel)](): Higher is better; unit: PERCENT. - [AP@0.15 (shower curtain)](): Higher is better; unit: PERCENT. - [AP@0.15 (box)](): Higher is better; unit: PERCENT. - [AP@0.15 (white board)](): Higher is better; unit: PERCENT. - [AP@0.15 (person)](): Higher is better; unit: PERCENT. - [AP@0.15 (night stand)](): Higher is better; unit: PERCENT. - [AP@0.15 (toilet)](): Higher is better; unit: PERCENT. - [AP@0.15 (sink)](): Higher is better; unit: PERCENT. - [AP@0.15 (lamp)](): Higher is better; unit: PERCENT. - [AP@0.15 (bathtub)](): Higher is better; unit: PERCENT. - [AP@0.15 (bag)](): Higher is better; unit: PERCENT. - [AP@0.15 (wall)](): Higher is better; unit: PERCENT. - [AP@0.15 (floor)](): Higher is better; unit: PERCENT. - [AP@0.15 (ceiling)](): Higher is better; unit: PERCENT. ## Recorded results | Method | AP@0.15 (Overall) | Result date | Report paper | | --- | --- | --- | --- | | 3DiffTection | 19.01 | Not recorded | 68fa711e1e209131cea4f368 | | 3DiffTection | 16.68 | Not recorded | 68fa711e1e209131cea4f368 | | DIFT-SD | 15.35 | Not recorded | 68fa711e1e209131cea4f368 | | CubeRCNN | 14.68 | Not recorded | 68fa711e1e209131cea4f368 | ### 3DiffTection AP@0.15 (Overall): 19.01. Result ID: 6991b2643a6c31484909fbf9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"StableDiff+Geo-Ctr","Pre-training Data":"LION5B & Aktscn","Pre-training Task":"Generation","Trained Module":"Sem-Ctr+3D Head"}. This record does not assert independent reproduction. ### 3DiffTection AP@0.15 (Overall): 16.68. Result ID: 6991b2643a6c31484909fbf8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"StableDiff+Geo-Ctr","Pre-training Data":"LION5B & Aktscn","Pre-training Task":"Generation","Trained Module":"3D Head"}. This record does not assert independent reproduction. ### DIFT-SD AP@0.15 (Overall): 15.35. Result ID: 6991b2643a6c31484909fbf6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"StableDiff","Pre-training Data":"LION5B (Schuhmann et al., 2022)","Pre-training Task":"Generation","Trained Module":"3D Head"}. This record does not assert independent reproduction. ### CubeRCNN AP@0.15 (Overall): 14.68. Result ID: 6991b2643a6c31484909fbf7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"DLA34","Pre-training Data":"ImageNet & Aktscn","Pre-training Task":"Classification & 3D det.","Trained Module":"3D Head"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.