# Zero-shot Part Segmentation on Pascal-Part-116 cross-dataset (trained on PartImageNet) (test)

Source: [Zero-shot Part Segmentation on Pascal-Part-116 cross-dataset (trained on PartImageNet) (test)](<https://www.sota2.com/research/sota/zero-shot-part-segmentation-on-pascal-part-116-cross-dataset-trained-on-partimagenet-test>)

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

Data updated: 2026-02-26T06:10:57.548Z

Evaluation of Zero-shot Part Segmentation performance on the Pascal-Part-116 subset, specifically when the model was trained on PartImageNet. Results report mean IoU metrics for both predicted parts (Pred-All) and oracle object alignments (Oracle-Obj).

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 | Zero-shot Part Segmentation |
| Dataset (including any stated split) | Pascal-Part-116 cross-dataset (trained on PartImageNet) (test) |
| Metric | mIoU (Pred-All) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [mIoU (Pred-All)](<https://www.sota2.com/research/sota/zero-shot-part-segmentation-on-pascal-part-116-cross-dataset-trained-on-partimagenet-test.md?metric=mIoU%20(Pred-All)>): Higher is better; unit: PERCENT.

- [mIoU (Oracle-Obj)](<https://www.sota2.com/research/sota/zero-shot-part-segmentation-on-pascal-part-116-cross-dataset-trained-on-partimagenet-test.md?metric=mIoU%20(Oracle-Obj)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | mIoU (Pred-All) | Result date | Report paper |
| --- | --- | --- | --- |
| PartCLIPSeg | 14.74 | Not recorded | 68fa76101e209131cea6b388 |
| CLIPSeg | 11.72 | Not recorded | 68fa76101e209131cea6b388 |

### PartCLIPSeg

mIoU (Pred-All): 14.74. Result ID: 6994238f3a6c3148491b5394.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76101e209131cea6b388.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76101e209131cea6b388.md>)

[Implementation](<https://github.com/kaist-cvml-lab/part-clipseg>)

Recorded evaluation configuration: {"Backbone":"ViT-B/16","Source Dataset":"PartImageNet"}. This record does not assert independent reproduction.

### CLIPSeg

mIoU (Pred-All): 11.72. Result ID: 6994238f3a6c3148491b5393.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76101e209131cea6b388.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c2f1e209131cea32a66.md>)

[Implementation](<https://eckerlab.org/code/clipseg>)

Recorded evaluation configuration: {"Backbone":"ViT-B/16","Source Dataset":"PartImageNet"}. This record does not assert independent reproduction.

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