# 3D Object Detection on SUN RGB-D V2 (test)

Source: [3D Object Detection on SUN RGB-D V2 (test)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:23:09.873Z

Evaluation of 3D object detection performance on the test split of the SUN RGB-D V2 dataset, reporting class-specific metrics (e.g., bed, table, sofa) and overall mAP scores.

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 V2 (test) |
| Metric | AP (bed) |
| 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

- [AP (bed)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(bed)>): Higher is better; unit: PERCENT.

- [AP (table)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(table)>): Higher is better; unit: PERCENT.

- [AP (sofa)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(sofa)>): Higher is better; unit: PERCENT.

- [AP (chair)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(chair)>): Higher is better; unit: PERCENT.

- [AP (toilet)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(toilet)>): Higher is better; unit: PERCENT.

- [AP (desk)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(desk)>): Higher is better; unit: PERCENT.

- [AP (dresser)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(dresser)>): Higher is better; unit: PERCENT.

- [AP (night-stand)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(night-stand)>): Higher is better; unit: PERCENT.

- [AP (book-shelf)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(book-shelf)>): Higher is better; unit: PERCENT.

- [AP (bathtub)](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=AP%20(bathtub)>): Higher is better; unit: PERCENT.

- [mAP](<https://www.sota2.com/research/sota/3d-object-detection-on-sun-rgb-d-v2-test.md?metric=mAP>): Higher is better; unit: PERCENT.

## Recorded results

| Method | AP (bed) | Result date | Report paper |
| --- | --- | --- | --- |
| RS-CNN | 85.7 | Not recorded | 68fa69bc1e209131cea23e01 |
| PointNet++ | 84.8 | Not recorded | 68fa69bc1e209131cea23e01 |
| Geometric Back-projection Network | 84.8 | Not recorded | 68fa69bc1e209131cea23e01 |
| DGCNN | 84.6 | Not recorded | 68fa69bc1e209131cea23e01 |
| PointNet | 46.2 | Not recorded | 68fa69bc1e209131cea23e01 |

### RS-CNN

AP (bed): 85.7. Result ID: 698ddfaaffc3901de0994c70.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65c41e209131cea0c510.md>)

[Implementation](<https://github.com/Yochengliu/Relation-Shape-CNN>)

Recorded evaluation configuration: {"backbone":"RS-CNN","structure":"VoteNet"}. This record does not assert independent reproduction.

### PointNet++

AP (bed): 84.8. Result ID: 698ddfaaffc3901de0994c6e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63641e209131ce9fe5a5.md>)

Recorded evaluation configuration: {"backbone":"PointNet++","structure":"VoteNet"}. This record does not assert independent reproduction.

### Geometric Back-projection Network

AP (bed): 84.8. Result ID: 698ddfaaffc3901de0994c71.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Implementation](<https://github.com/ShiQiu0419/GBNet>)

Recorded evaluation configuration: {"backbone":"GBNet","structure":"VoteNet"}. This record does not assert independent reproduction.

### DGCNN

AP (bed): 84.6. Result ID: 698ddfaaffc3901de0994c6f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65e11e209131cea0d1ac.md>)

[Implementation](<https://github.com/WangYueFt/dgcnn>)

Recorded evaluation configuration: {"backbone":"DGCNN","structure":"VoteNet"}. This record does not assert independent reproduction.

### PointNet

AP (bed): 46.2. Result ID: 698ddfaaffc3901de0994c6d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e01.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa634f1e209131ce9fe019.md>)

Recorded evaluation configuration: {"backbone":"PointNet","structure":"VoteNet"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.