# 3D Object Detection on Waymo Open Dataset Domain Adaptation from nuScenes and Lyft (val)

Source: [3D Object Detection on Waymo Open Dataset Domain Adaptation from nuScenes and Lyft (val)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val>)

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

Data updated: 2026-05-12T14:22:30.814Z

Evaluation of 3D Object Detection performance on the validation split of the Waymo Open Dataset Domain Adaptation subset (derived from nuScenes and Lyft), reporting L1/L2 mAP and mAPH metrics across Car, Pedestrian, and Cyclist classes.

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 Dataset Domain Adaptation from nuScenes and Lyft (val) |
| Metric | mAP L1 (Car) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [mAP L1 (Car)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L1%20(Car)>): Higher is better; unit: PERCENT.

- [mAP L2 (Car)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L2%20(Car)>): Higher is better; unit: PERCENT.

- [mAP L1 (Pedestrian)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L1%20(Pedestrian)>): Higher is better; unit: PERCENT.

- [mAP L2 (Pedestrian)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L2%20(Pedestrian)>): Higher is better; unit: PERCENT.

- [mAP L1 (Cyclist)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L1%20(Cyclist)>): Higher is better; unit: PERCENT.

- [mAP L2 (Cyclist)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAP%20L2%20(Cyclist)>): Higher is better; unit: PERCENT.

- [mAPH L1 (Car)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L1%20(Car)>): Higher is better; unit: PERCENT.

- [mAPH L2 (Car)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L2%20(Car)>): Higher is better; unit: PERCENT.

- [mAPH L1 (Pedestrian)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L1%20(Pedestrian)>): Higher is better; unit: PERCENT.

- [mAPH L2 (Pedestrian)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L2%20(Pedestrian)>): Higher is better; unit: PERCENT.

- [mAPH L1 (Cyclist)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L1%20(Cyclist)>): Higher is better; unit: PERCENT.

- [mAPH L2 (Cyclist)](<https://www.sota2.com/research/sota/3d-object-detection-on-waymo-open-dataset-domain-adaptation-from-nuscenes-and-lyft-val.md?metric=mAPH%20L2%20(Cyclist)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | mAP L1 (Car) | Result date | Report paper |
| --- | --- | --- | --- |
| Oracle | 78.8 | Not recorded | 6a02b494dbc836b87d824d67 |
| HSC-DC | 68.9 | Not recorded | 6a02b494dbc836b87d824d67 |
| HSC-DC + PGW-MF | 68.2 | Not recorded | 6a02b494dbc836b87d824d67 |
| Pseudo label | 67.4 | Not recorded | 6a02b494dbc836b87d824d67 |
| Disentangle learning | 66.9 | Not recorded | 6a02b494dbc836b87d824d67 |
| Source Only (w/ PTDA) | 63.8 | Not recorded | 6a02b494dbc836b87d824d67 |

### Oracle

mAP L1 (Car): 78.8. Result ID: 6a0336cc47f34ce924e7fb16.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

Recorded evaluation configuration: {"Supervision":"Fully Supervised Target"}. This record does not assert independent reproduction.

### HSC-DC

mAP L1 (Car): 68.9. Result ID: 6a0336cc47f34ce924e7fb14.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

Recorded evaluation configuration: {"Module":"Hierarchical Spatially-Conditioned Domain Classifiers"}. This record does not assert independent reproduction.

### HSC-DC + PGW-MF

mAP L1 (Car): 68.2. Result ID: 6a0336cc47f34ce924e7fb15.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

Recorded evaluation configuration: {"Module":"HSC-DC and Prototype Graph Weighted Multi-source Fusion"}. This record does not assert independent reproduction.

### Pseudo label

mAP L1 (Car): 67.4. Result ID: 6a0336cc47f34ce924e7fb12.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64301e209131cea03556.md>)

[Implementation](<https://naoto0804.github.io/cross_domain_detection>)

Recorded evaluation configuration: {"Strategy":"Pseudo labeling"}. This record does not assert independent reproduction.

### Disentangle learning

mAP L1 (Car): 66.9. Result ID: 6a0336cc47f34ce924e7fb13.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

Recorded evaluation configuration: {"Strategy":"Disentangle learning"}. This record does not assert independent reproduction.

### Source Only (w/ PTDA)

mAP L1 (Car): 63.8. Result ID: 6a0336cc47f34ce924e7fb11.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b494dbc836b87d824d67.md>)

Recorded evaluation configuration: {"PTDA":"true"}. This record does not assert independent reproduction.

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