# Semantic Segmentation on SemanticKITTI (Per-Class IoU and Stuff Metrics)

Source: [Semantic Segmentation on SemanticKITTI (Per-Class IoU and Stuff Metrics)](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics>)

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

Data updated: 2026-02-26T05:38:56.901Z

Evaluation of Semantic Segmentation performance on the SemanticKITTI dataset, reporting mean IoU (mIoU), instance mIoU (mIoUInst), stuff mIoU (mIoUStuff), and detailed per-class Intersection over Union 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 | Semantic Segmentation |
| Dataset (including any stated split) | SemanticKITTI |
| Metric | Car IoU |
| 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

- [Car IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Car%20IoU>): Higher is better; unit: PERCENT.

- [Bicycle IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Bicycle%20IoU>): Higher is better; unit: PERCENT.

- [Motorcycle IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Motorcycle%20IoU>): Higher is better; unit: PERCENT.

- [Truck IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Truck%20IoU>): Higher is better; unit: PERCENT.

- [Other Vehicle IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Other%20Vehicle%20IoU>): Higher is better; unit: PERCENT.

- [Person IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Person%20IoU>): Higher is better; unit: PERCENT.

- [Bicyclist IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Bicyclist%20IoU>): Higher is better; unit: PERCENT.

- [Motorcyclist IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Motorcyclist%20IoU>): Higher is better; unit: PERCENT.

- [Road IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Road%20IoU>): Higher is better; unit: PERCENT.

- [Parking IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Parking%20IoU>): Higher is better; unit: PERCENT.

- [Sidewalk IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Sidewalk%20IoU>): Higher is better; unit: PERCENT.

- [Other Ground IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Other%20Ground%20IoU>): Higher is better; unit: PERCENT.

- [Building IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Building%20IoU>): Higher is better; unit: PERCENT.

- [Fence IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Fence%20IoU>): Higher is better; unit: PERCENT.

- [Vegetation IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Vegetation%20IoU>): Higher is better; unit: PERCENT.

- [Trunk IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Trunk%20IoU>): Higher is better; unit: PERCENT.

- [Terrain IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Terrain%20IoU>): Higher is better; unit: PERCENT.

- [Pole IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Pole%20IoU>): Higher is better; unit: PERCENT.

- [Traffic Sign IoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=Traffic%20Sign%20IoU>): Higher is better; unit: PERCENT.

- [mIoU (Mean IoU)](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=mIoU%20(Mean%20IoU)>): Higher is better; unit: PERCENT.

- [mIoUInst (Instance mIoU)](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=mIoUInst%20(Instance%20mIoU)>): Higher is better; unit: PERCENT.

- [mIoUStuff (Stuff mIoU)](<https://www.sota2.com/research/sota/semantic-segmentation-on-semantickitti-per-class-iou-and-stuff-metrics.md?metric=mIoUStuff%20(Stuff%20mIoU)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Car IoU | Result date | Report paper |
| --- | --- | --- | --- |
| CAP | 95.8 | Not recorded | 697846cea2b28e761d87c74d |
| KPRNet | 95.5 | Not recorded | 697846cea2b28e761d87c74d |
| RangeViT | 95.4 | Not recorded | 697846cea2b28e761d87c74d |
| CWAP | 93.3 | Not recorded | 697846cea2b28e761d87c74d |
| SqueezeSegV3 | 92.5 | Not recorded | 697846cea2b28e761d87c74d |
| SalsaNext | 91.9 | Not recorded | 697846cea2b28e761d87c74d |
| RangeNet++ | 91.4 | Not recorded | 697846cea2b28e761d87c74d |

### CAP

Car IoU: 95.8. Result ID: 69978d495698665ec4e272bc.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"Centerness-Aware Projection"}. This record does not assert independent reproduction.

### KPRNet

Car IoU: 95.5. Result ID: 69978d495698665ec4e272ba.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68341e209131cea1abf9.md>)

[Implementation](<https://github.com/DeyvidKochanov-TomTom/kprnet>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"KPRNet projection"}. This record does not assert independent reproduction.

### RangeViT

Car IoU: 95.4. Result ID: 69978d495698665ec4e272bb.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f211e209131cea43e64.md>)

[Implementation](<https://github.com/valeoai/rangevit>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"Depth-based projection"}. This record does not assert independent reproduction.

### CWAP

Car IoU: 93.3. Result ID: 69978d495698665ec4e272bd.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"Class-Weighted-Aware Projection"}. This record does not assert independent reproduction.

### SqueezeSegV3

Car IoU: 92.5. Result ID: 69978d495698665ec4e272b8.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69bc1e209131cea23e0d.md>)

[Implementation](<https://github.com/chenfengxu714/SqueezeSegV3>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"SqueezeSegV3 projection"}. This record does not assert independent reproduction.

### SalsaNext

Car IoU: 91.9. Result ID: 69978d495698665ec4e272b9.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"SalsaNext projection"}. This record does not assert independent reproduction.

### RangeNet++

Car IoU: 91.4. Result ID: 69978d495698665ec4e272b7.

[Reporting paper](<https://www.sota2.com/research/paper/697846cea2b28e761d87c74d.md>)

Recorded evaluation configuration: {"Architecture":"RangeViT","Projection Strategy":"RangeNet++ projection"}. This record does not assert independent reproduction.

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