# Traffic Sign Captioning on TrafficSignEval

Source: [Traffic Sign Captioning on TrafficSignEval](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval>)

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

Data updated: 2026-04-14T09:11:06.309Z

Evaluation of Traffic Sign Captioning performance on the TrafficSignEval dataset, reporting attribute-specific metrics such as Direction, Notice, Lane, and Consistency (Const.).

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 | Traffic Sign Captioning |
| Dataset (including any stated split) | TrafficSignEval |
| Metric | Direction |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 11 of 11 |
| 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

- [Direction](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval.md?metric=Direction>): Higher is better; unit: PERCENT.

- [Notice](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval.md?metric=Notice>): Higher is better; unit: PERCENT.

- [Lane](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval.md?metric=Lane>): Higher is better; unit: PERCENT.

- [Consistency](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval.md?metric=Consistency>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/traffic-sign-captioning-on-trafficsigneval.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Direction | Result date | Report paper |
| --- | --- | --- | --- |
| SignReasoner | 80.01 | Not recorded | 69ddca3e7d0172140ec79c1d |
| SignReasoner | 79.41 | Not recorded | 69ddca3e7d0172140ec79c1d |
| SignReasoner | 71 | Not recorded | 69ddca3e7d0172140ec79c1d |
| Pangu-MM-7B | 70.59 | Not recorded | 69ddca3e7d0172140ec79c1d |
| SignReasoner | 70.59 | Not recorded | 69ddca3e7d0172140ec79c1d |
| Qwen2.5-VL-72B | 64.71 | Not recorded | 69ddca3e7d0172140ec79c1d |
| Qwen2.5-VL-7B | 58.82 | Not recorded | 69ddca3e7d0172140ec79c1d |
| InternVL3-8B | 55.88 | Not recorded | 69ddca3e7d0172140ec79c1d |
| InternVL3-78B | 55.88 | Not recorded | 69ddca3e7d0172140ec79c1d |
| Pangu-MM-7B | 55.88 | Not recorded | 69ddca3e7d0172140ec79c1d |
| Qwen2.5-VL-7B | 47.06 | Not recorded | 69ddca3e7d0172140ec79c1d |

### SignReasoner

Direction: 80.01. Result ID: 69de049647f34ce924c5a39b.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Base Model":"Pangu-MM-7B","Performance Protocol":"FSU-GRPO Training"}. This record does not assert independent reproduction.

### SignReasoner

Direction: 79.41. Result ID: 69de049647f34ce924c5a399.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Base Model":"Pangu-MM-7B","Performance Protocol":"Caption-FSU Distillation"}. This record does not assert independent reproduction.

### SignReasoner

Direction: 71. Result ID: 69de049647f34ce924c5a39a.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-VL-7B","Performance Protocol":"FSU-GRPO Training"}. This record does not assert independent reproduction.

### Pangu-MM-7B

Direction: 70.59. Result ID: 69de049647f34ce924c5a397.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Performance Protocol":"Caption SFT"}. This record does not assert independent reproduction.

### SignReasoner

Direction: 70.59. Result ID: 69de049647f34ce924c5a398.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-VL-7B","Performance Protocol":"Caption-FSU Distillation"}. This record does not assert independent reproduction.

### Qwen2.5-VL-72B

Direction: 64.71. Result ID: 69de049647f34ce924c5a394.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Performance Protocol":"Zero-Shot"}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B

Direction: 58.82. Result ID: 69de049647f34ce924c5a396.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Performance Protocol":"Caption SFT"}. This record does not assert independent reproduction.

### InternVL3-8B

Direction: 55.88. Result ID: 69de049647f34ce924c5a391.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76fa5.md>)

Recorded evaluation configuration: {"Performance Protocol":"Zero-Shot"}. This record does not assert independent reproduction.

### InternVL3-78B

Direction: 55.88. Result ID: 69de049647f34ce924c5a392.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76fa5.md>)

Recorded evaluation configuration: {"Performance Protocol":"Zero-Shot"}. This record does not assert independent reproduction.

### Pangu-MM-7B

Direction: 55.88. Result ID: 69de049647f34ce924c5a395.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

Recorded evaluation configuration: {"Performance Protocol":"Zero-Shot"}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B

Direction: 47.06. Result ID: 69de049647f34ce924c5a393.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca3e7d0172140ec79c1d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Performance Protocol":"Zero-Shot"}. This record does not assert independent reproduction.

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