# Part Segmentation on Scan2CAD (test)

Source: [Part Segmentation on Scan2CAD (test)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test>)

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

Data updated: 2026-02-26T05:38:10.709Z

Evaluation of Part Segmentation performance on the test split of the Scan2CAD dataset, reporting Chamfer Distance and IoU metrics across various object categories.

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 | Part Segmentation |
| Dataset (including any stated split) | Scan2CAD official (test) |
| Metric | CD (chair) |
| Unit | SCALAR |
| Direction | Lower 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

- [CD (chair)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(chair)>): Lower is better; unit: SCALAR.

- [CD (table)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(table)>): Lower is better; unit: SCALAR.

- [CD (cab.)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(cab.)>): Lower is better; unit: SCALAR.

- [CD (bkshlf)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(bkshlf)>): Lower is better; unit: SCALAR.

- [CD (bed)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(bed)>): Lower is better; unit: SCALAR.

- [CD (bin)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(bin)>): Lower is better; unit: SCALAR.

- [CD (class avg)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(class%20avg)>): Lower is better; unit: SCALAR.

- [CD (inst avg)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=CD%20(inst%20avg)>): Lower is better; unit: SCALAR.

- [IoU (chair)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(chair)>): Higher is better; unit: PERCENT.

- [IoU (table)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(table)>): Higher is better; unit: PERCENT.

- [IoU (cab.)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(cab.)>): Higher is better; unit: PERCENT.

- [IoU (bkshlf)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(bkshlf)>): Higher is better; unit: PERCENT.

- [IoU (bed)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(bed)>): Higher is better; unit: PERCENT.

- [IoU (bin)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(bin)>): Higher is better; unit: PERCENT.

- [IoU (class avg)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(class%20avg)>): Higher is better; unit: PERCENT.

- [IoU (inst avg)](<https://www.sota2.com/research/sota/part-segmentation-on-scan2cad-test.md?metric=IoU%20(inst%20avg)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | CD (chair) | Result date | Report paper |
| --- | --- | --- | --- |
| Neural Part Priors | 0.021 | Not recorded | 68fa6ee71e209131cea42c73 |
| MLCVNet + StructureNet | 0.025 | Not recorded | 68fa6ee71e209131cea42c73 |
| Bokhovkin et al. | 0.027 | Not recorded | 68fa6ee71e209131cea42c73 |
| SG-NN + MLCVNet + PointGroup | 0.056 | Not recorded | 68fa6ee71e209131cea42c73 |

### Neural Part Priors

CD (chair): 0.021. Result ID: 69906c0b3a6c31484900713a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ee71e209131cea42c73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ee71e209131cea42c73.md>)

[Implementation](<https://www.sota2.com/alexeybokhovkin.github.io/neural-part-priors/>)

Recorded evaluation configuration: {"Optimizer":"Adam","Learning Rate":"3e-4/3e-5","Iterations":600}. This record does not assert independent reproduction.

### MLCVNet + StructureNet

CD (chair): 0.025. Result ID: 69906c0b3a6c314849007138.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ee71e209131cea42c73.md>)

Recorded evaluation configuration: {"Detector":"MLCVNet","Part Priors":"StructureNet"}. This record does not assert independent reproduction.

### Bokhovkin et al.

CD (chair): 0.027. Result ID: 69906c0b3a6c314849007139.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ee71e209131cea42c73.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68db1e209131cea1eb42.md>)

Recorded evaluation configuration: {"Method Type":"Coarse Part Decomposition"}. This record does not assert independent reproduction.

### SG-NN + MLCVNet + PointGroup

CD (chair): 0.056. Result ID: 69906c0b3a6c314849007137.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6ee71e209131cea42c73.md>)

Recorded evaluation configuration: {"Detector":"MLCVNet","Segmentor":"PointGroup","Geometric Association":"SG-NN"}. This record does not assert independent reproduction.

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