# Reasoning Segmentation on GeoSeg-Bench

Source: [Reasoning Segmentation on GeoSeg-Bench](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench>)

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

Data updated: 2026-03-05T09:11:09.363Z

Evaluation of Reasoning Segmentation performance on the GeoSeg-Bench dataset, reporting standard segmentation and boundary metrics including IoU, Dice, Accuracy, Precision, Recall, Specificity, and Boundary F-score (BF).

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 | Reasoning Segmentation |
| Dataset (including any stated split) | GeoSeg-Bench |
| Metric | IoU |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

- [IoU](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=IoU>): Higher is better; unit: PERCENT.

- [Dice](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Dice>): Higher is better; unit: PERCENT.

- [Accuracy](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Precision](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Precision>): Higher is better; unit: PERCENT.

- [Recall](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Recall>): Higher is better; unit: PERCENT.

- [Specificity](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Specificity>): Higher is better; unit: PERCENT.

- [Boundary F-score (BF)](<https://www.sota2.com/research/sota/reasoning-segmentation-on-geoseg-bench.md?metric=Boundary%20F-score%20(BF)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | IoU | Result date | Report paper |
| --- | --- | --- | --- |
| GeoSeg | 56.4 | Not recorded | 69a90e162d603e6de3e408ee |
| LISA-7B | 39.5 | Not recorded | 69a90e162d603e6de3e408ee |
| CLIP Surgery + SAM3 | 24.7 | Not recorded | 69a90e162d603e6de3e408ee |
| SAM3 | 19.9 | Not recorded | 69a90e162d603e6de3e408ee |
| Grounded SAM | 18.6 | Not recorded | 69a90e162d603e6de3e408ee |
| LISA++ | 12.4 | Not recorded | 69a90e162d603e6de3e408ee |
| BAGEL | 10.7 | Not recorded | 69a90e162d603e6de3e408ee |
| GSVA | 9 | Not recorded | 69a90e162d603e6de3e408ee |
| UniWorld | 8.8 | Not recorded | 69a90e162d603e6de3e408ee |
| Qwen-Edit | 8.7 | Not recorded | 69a90e162d603e6de3e408ee |
| OmniGen2 | 6.8 | Not recorded | 69a90e162d603e6de3e408ee |
| Hunyuan | 5.6 | Not recorded | 69a90e162d603e6de3e408ee |
| GLAMM | 0 | Not recorded | 69a90e162d603e6de3e408ee |
| PixelLM | 0 | Not recorded | 69a90e162d603e6de3e408ee |

### GeoSeg

IoU: 56.4. Result ID: 69a948966dbb7075b952f7cc.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

Recorded evaluation configuration: {"Method Category":"Ours","FPS":"0.221"}. This record does not assert independent reproduction.

### LISA-7B

IoU: 39.5. Result ID: 69a948966dbb7075b952f7c4.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730a1e209131cea5a0b2.md>)

[Implementation](<https://www.sota2.com/github.com/dvlab-research/LISA>)

Recorded evaluation configuration: {"Method Category":"Reasoning segmentation","FPS":"0.392"}. This record does not assert independent reproduction.

### CLIP Surgery + SAM3

IoU: 24.7. Result ID: 69a948966dbb7075b952f7c1.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

Recorded evaluation configuration: {"Method Category":"Generalist segmentation"}. This record does not assert independent reproduction.

### SAM3

IoU: 19.9. Result ID: 69a948966dbb7075b952f7bf.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6923e69bd6b0dd334292f0ad.md>)

[Implementation](<https://github.com/facebookresearch/sam3>)

Recorded evaluation configuration: {"Method Category":"Generalist segmentation"}. This record does not assert independent reproduction.

### Grounded SAM

IoU: 18.6. Result ID: 69a948966dbb7075b952f7c0.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71f11e209131cea53ab0.md>)

Recorded evaluation configuration: {"Method Category":"Generalist segmentation"}. This record does not assert independent reproduction.

### LISA++

IoU: 12.4. Result ID: 69a948966dbb7075b952f7c5.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71f41e209131cea53c42.md>)

Recorded evaluation configuration: {"Method Category":"Reasoning segmentation"}. This record does not assert independent reproduction.

### BAGEL

IoU: 10.7. Result ID: 69a948966dbb7075b952f7c9.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a311e209131cea81aad.md>)

[Implementation](<https://bagel-ai.org/>)

Recorded evaluation configuration: {"Method Category":"Open-source MLLMs","FPS":"0.001"}. This record does not assert independent reproduction.

### GSVA

IoU: 9. Result ID: 69a948966dbb7075b952f7c3.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa729f1e209131cea5799e.md>)

[Implementation](<https://github.com/LeapLabTHU/GSVA>)

Recorded evaluation configuration: {"Method Category":"Reasoning segmentation"}. This record does not assert independent reproduction.

### UniWorld

IoU: 8.8. Result ID: 69a948966dbb7075b952f7ca.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79831e209131cea7dee8.md>)

Recorded evaluation configuration: {"Method Category":"Open-source MLLMs","FPS":"0.015"}. This record does not assert independent reproduction.

### Qwen-Edit

IoU: 8.7. Result ID: 69a948966dbb7075b952f7c7.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a6a1e209131cea83137.md>)

Recorded evaluation configuration: {"Method Category":"Open-source MLLMs","FPS":"0.011"}. This record does not assert independent reproduction.

### OmniGen2

IoU: 6.8. Result ID: 69a948966dbb7075b952f7c8.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ba31e209131cea89864.md>)

Recorded evaluation configuration: {"Method Category":"Open-source MLLMs","FPS":"0.011"}. This record does not assert independent reproduction.

### Hunyuan

IoU: 5.6. Result ID: 69a948966dbb7075b952f7cb.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bb51e209131cea8a0c1.md>)

Recorded evaluation configuration: {"Method Category":"Open-source MLLMs","FPS":"0.002"}. This record does not assert independent reproduction.

### GLAMM

IoU: 0. Result ID: 69a948966dbb7075b952f7c2.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c2d8.md>)

[Implementation](<https://mbzuai-oryx.github.io/groundingLMM>)

Recorded evaluation configuration: {"Method Category":"Reasoning segmentation"}. This record does not assert independent reproduction.

### PixelLM

IoU: 0. Result ID: 69a948966dbb7075b952f7c6.

[Reporting paper](<https://www.sota2.com/research/paper/69a90e162d603e6de3e408ee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74241e209131cea5ff3b.md>)

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

Recorded evaluation configuration: {"Method Category":"Reasoning segmentation","FPS":"0.175"}. This record does not assert independent reproduction.

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