# 3D Object Detection on KITTI (val) (Car R40 AP)

Source: [3D Object Detection on KITTI (val) (Car R40 AP)](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap>)

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

Data updated: 2026-02-26T05:21:36.975Z

Evaluation of 3D Object Detection performance on the KITTI validation set, focusing specifically on Average Precision (AP) metrics derived from the R40 range evaluation for the Car class, covering both 3D and Bird's Eye View (BEV) results.

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) | KITTI (val) |
| Metric | 3D AP (R40) Easy |
| 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

- [3D AP (R40) Easy](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=3D%20AP%20(R40)%20Easy>): Higher is better; unit: PERCENT.

- [3D AP (R40) Moderate](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=3D%20AP%20(R40)%20Moderate>): Higher is better; unit: PERCENT.

- [3D AP (R40) Hard](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=3D%20AP%20(R40)%20Hard>): Higher is better; unit: PERCENT.

- [BEV AP (R40) Easy](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=BEV%20AP%20(R40)%20Easy>): Higher is better; unit: PERCENT.

- [BEV AP (R40) Moderate](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=BEV%20AP%20(R40)%20Moderate>): Higher is better; unit: PERCENT.

- [BEV AP (R40) Hard](<https://www.sota2.com/research/sota/3d-object-detection-on-kitti-val-car-r40-ap.md?metric=BEV%20AP%20(R40)%20Hard>): Higher is better; unit: PERCENT.

## Recorded results

| Method | 3D AP (R40) Easy | Result date | Report paper |
| --- | --- | --- | --- |
| CPDet3D | 94.1 | Not recorded | 68fa79591e209131cea7d33b |
| Voxel-RCNN | 92.3 | Not recorded | 68fa79591e209131cea7d33b |
| CoIn++ | 92 | Not recorded | 68fa79591e209131cea7d33b |
| CenterPoint | 89 | Not recorded | 68fa79591e209131cea7d33b |
| SS3D\* | 88.8 | Not recorded | 68fa79591e209131cea7d33b |
| CoIn | 84.5 | Not recorded | 68fa79591e209131cea7d33b |
| Voxel-RCNN | 72.5 | Not recorded | 68fa79591e209131cea7d33b |

### CPDet3D

3D AP (R40) Easy: 94.1. Result ID: 6995db603badfaffe6e1e5cb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

[Implementation](<https://github.com/zyrant/CPDet3D>)

Recorded evaluation configuration: {"Paradigm":"Sparse Supervised","Base Detector":"Voxel-RCNN","Annotated Cost":"2%"}. This record does not assert independent reproduction.

### Voxel-RCNN

3D AP (R40) Easy: 92.3. Result ID: 6995db603badfaffe6e1e5c6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa693b1e209131cea21032.md>)

[Implementation](<https://github.com/djiajunustc/Voxel-R-CNN>)

Recorded evaluation configuration: {"Paradigm":"Fully Supervised","Base Detector":"Voxel-RCNN","Present at":"AAAI"}. This record does not assert independent reproduction.

### CoIn++

3D AP (R40) Easy: 92. Result ID: 6995db603badfaffe6e1e5ca.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

Recorded evaluation configuration: {"Paradigm":"Sparse Supervised","Base Detector":"Voxel-RCNN","Present at":"ICCV","Annotated Cost":"2%"}. This record does not assert independent reproduction.

### CenterPoint

3D AP (R40) Easy: 89. Result ID: 6995db603badfaffe6e1e5c5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6af91e209131cea2b72c.md>)

[Implementation](<https://github.com/tianweiy/CenterPoint>)

Recorded evaluation configuration: {"Paradigm":"Fully Supervised","Base Detector":"Voxel-RCNN","Present at":"CVPR"}. This record does not assert independent reproduction.

### SS3D\*

3D AP (R40) Easy: 88.8. Result ID: 6995db603badfaffe6e1e5c8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

Recorded evaluation configuration: {"Paradigm":"Sparse Supervised","Base Detector":"Voxel-RCNN","Present at":"CVPR","Annotated Cost":"3%","Evaluation Protocol":"R11"}. This record does not assert independent reproduction.

### CoIn

3D AP (R40) Easy: 84.5. Result ID: 6995db603badfaffe6e1e5c9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

Recorded evaluation configuration: {"Paradigm":"Sparse Supervised","Base Detector":"Voxel-RCNN","Present at":"ICCV","Annotated Cost":"2%"}. This record does not assert independent reproduction.

### Voxel-RCNN

3D AP (R40) Easy: 72.5. Result ID: 6995db603badfaffe6e1e5c7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d33b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa693b1e209131cea21032.md>)

[Implementation](<https://github.com/djiajunustc/Voxel-R-CNN>)

Recorded evaluation configuration: {"Paradigm":"Sparse Supervised","Base Detector":"Voxel-RCNN","Present at":"AAAI","Annotated Cost":"2%"}. This record does not assert independent reproduction.

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