# 3D Object Detection on Waymo Open Level 2 (val, 100%)

Source: [3D Object Detection on Waymo Open Level 2 (val, 100%)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent>)

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

Data updated: 2026-02-26T06:08:11.813Z

Evaluation of 3D Object Detection performance on the validation split of the Waymo Open Dataset using the Level 2 difficulty configuration (100% data usage). Metrics include AP/APH for Vehicle, Pedestrian, and Cyclist classes, alongside inference latency.

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) | Waymo Open Level 2 100% train (val) |
| Metric | Vehicle AP |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Vehicle AP](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Vehicle%20AP>): Higher is better; unit: PERCENT.

- [Vehicle APH](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Vehicle%20APH>): Higher is better; unit: PERCENT.

- [Pedestrian AP](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Pedestrian%20AP>): Higher is better; unit: PERCENT.

- [Pedestrian APH](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Pedestrian%20APH>): Higher is better; unit: PERCENT.

- [Cyclist AP](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Cyclist%20AP>): Higher is better; unit: PERCENT.

- [Cyclist APH](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Cyclist%20APH>): Higher is better; unit: PERCENT.

- [Latency (ms)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-level-2-val-100-percent.md?metric=Latency%20(ms)>): Lower is better; unit: TIME.

## Recorded results

| Method | Vehicle AP | Result date | Report paper |
| --- | --- | --- | --- |
| FSHNetbase | 74.5 | Not recorded | 68fa79131e209131cea7c001 |
| FSHNetlight | 73 | Not recorded | 68fa79131e209131cea7c001 |
| SAFDNet | 72.7 | Not recorded | 68fa79131e209131cea7c001 |
| VoxelNext | 69.9 | Not recorded | 68fa79131e209131cea7c001 |

### FSHNetbase

Vehicle AP: 74.5. Result ID: 6995c2bc3badfaffe6c58fa8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Implementation](<https://github.com/Say2L/FSHNet>)

Recorded evaluation configuration: {"GPU":"RTX 3090","Training Data Scale":"100%"}. This record does not assert independent reproduction.

### FSHNetlight

Vehicle AP: 73. Result ID: 6995c2bc3badfaffe6c58fa7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Implementation](<https://github.com/Say2L/FSHNet>)

Recorded evaluation configuration: {"GPU":"RTX 3090","Training Data Scale":"100%"}. This record does not assert independent reproduction.

### SAFDNet

Vehicle AP: 72.7. Result ID: 6995c2bc3badfaffe6c58fa6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74e71e209131cea6499b.md>)

Recorded evaluation configuration: {"GPU":"RTX 3090","Training Data Scale":"100%"}. This record does not assert independent reproduction.

### VoxelNext

Vehicle AP: 69.9. Result ID: 6995c2bc3badfaffe6c58fa5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c001.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ed21e209131cea42682.md>)

[Implementation](<https://www.sota2.com/github.com/dvlab-research/VoxelNeXt>)

Recorded evaluation configuration: {"GPU":"RTX 3090","Training Data Scale":"100%"}. This record does not assert independent reproduction.

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