# Event-level Video Anomaly Detection on ShanghaiTech (SHT) 19 (test)

Source: [Event-level Video Anomaly Detection on ShanghaiTech (SHT) 19 (test)](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test>)

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

Data updated: 2026-04-13T09:30:12.366Z

Evaluation of Event-level Video Anomaly Detection performance on the test split of the ShanghaiTech (SHT) 19 dataset, measured using tIOU-based Precision, Recall, and F1 scores across various temporal thresholds.

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 | Event-level Video Anomaly Detection |
| Dataset (including any stated split) | ShanghaiTech (SHT) 19 (test) |
| Metric | Precision @ tIOU=0.5 |
| 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

- [Precision @ tIOU=0.5](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Precision%20%40%20tIOU%3D0.5>): Higher is better; unit: PERCENT.

- [Recall @ tIOU=0.5](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Recall%20%40%20tIOU%3D0.5>): Higher is better; unit: PERCENT.

- [F1 Score @ tIOU=0.5](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=F1%20Score%20%40%20tIOU%3D0.5>): Higher is better; unit: SCALAR.

- [Precision @ tIOU=0.4](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Precision%20%40%20tIOU%3D0.4>): Higher is better; unit: PERCENT.

- [Recall @ tIOU=0.4](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Recall%20%40%20tIOU%3D0.4>): Higher is better; unit: PERCENT.

- [F1 Score @ tIOU=0.4](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=F1%20Score%20%40%20tIOU%3D0.4>): Higher is better; unit: SCALAR.

- [Precision @ tIOU=0.3](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Precision%20%40%20tIOU%3D0.3>): Higher is better; unit: PERCENT.

- [Recall @ tIOU=0.3](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Recall%20%40%20tIOU%3D0.3>): Higher is better; unit: PERCENT.

- [F1 Score @ tIOU=0.3](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=F1%20Score%20%40%20tIOU%3D0.3>): Higher is better; unit: SCALAR.

- [Precision @ tIOU=0.2](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Precision%20%40%20tIOU%3D0.2>): Higher is better; unit: PERCENT.

- [Recall @ tIOU=0.2](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Recall%20%40%20tIOU%3D0.2>): Higher is better; unit: PERCENT.

- [F1 Score @ tIOU=0.2](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=F1%20Score%20%40%20tIOU%3D0.2>): Higher is better; unit: SCALAR.

- [Average F1 Score](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md?metric=Average%20F1%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Precision @ tIOU=0.5 | Result date | Report paper |
| --- | --- | --- | --- |
| STG-NF | 55.41 | Not recorded | 69dc78767d0172140ec78a4e |
| STG-NF | 50.59 | Not recorded | 69dc78767d0172140ec78a4e |
| SPARTA | 48.19 | Not recorded | 69dc78767d0172140ec78a4e |
| Dual (Ours) | 46.27 | Not recorded | 69dc78767d0172140ec78a4e |
| SPARTA | 44.62 | Not recorded | 69dc78767d0172140ec78a4e |
| TS-GAD | 41.38 | Not recorded | 69dc78767d0172140ec78a4e |
| TS-GAD | 36.79 | Not recorded | 69dc78767d0172140ec78a4e |

### STG-NF

Precision @ tIOU=0.5: 55.41. Result ID: 69dcb77747f34ce924c3f995.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa705c1e209131cea4a976.md>)

[Implementation](<https://github.com/orhir/STG-NF>)

Recorded evaluation configuration: {"Configuration":"Baselines + Proposed Post-Processing","Refinement":"w/ ES","Threshold":"τEER"}. This record does not assert independent reproduction.

### STG-NF

Precision @ tIOU=0.5: 50.59. Result ID: 69dcb77747f34ce924c3f992.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa705c1e209131cea4a976.md>)

[Implementation](<https://github.com/orhir/STG-NF>)

Recorded evaluation configuration: {"Configuration":"Baselines (frame grouping)","Threshold":"τEER"}. This record does not assert independent reproduction.

### SPARTA

Precision @ tIOU=0.5: 48.19. Result ID: 69dcb77747f34ce924c3f997.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777b1e209131cea737a2.md>)

Recorded evaluation configuration: {"Configuration":"Baselines + Proposed Post-Processing","Refinement":"w/ ES","Threshold":"τEER"}. This record does not assert independent reproduction.

### Dual (Ours)

Precision @ tIOU=0.5: 46.27. Result ID: 69dcb77747f34ce924c3f998.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

Recorded evaluation configuration: {"Configuration":"Dual-Branch Event Detection","Threshold":"τEER"}. This record does not assert independent reproduction.

### SPARTA

Precision @ tIOU=0.5: 44.62. Result ID: 69dcb77747f34ce924c3f994.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777b1e209131cea737a2.md>)

Recorded evaluation configuration: {"Configuration":"Baselines (frame grouping)","Threshold":"τEER"}. This record does not assert independent reproduction.

### TS-GAD

Precision @ tIOU=0.5: 41.38. Result ID: 69dcb77747f34ce924c3f996.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b71e209131cea5e17d.md>)

[Implementation](<https://github.com/TeCSAR-UNCC/TSGAD>)

Recorded evaluation configuration: {"Configuration":"Baselines + Proposed Post-Processing","Refinement":"w/ ES","Threshold":"τEER"}. This record does not assert independent reproduction.

### TS-GAD

Precision @ tIOU=0.5: 36.79. Result ID: 69dcb77747f34ce924c3f993.

[Reporting paper](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b71e209131cea5e17d.md>)

[Implementation](<https://github.com/TeCSAR-UNCC/TSGAD>)

Recorded evaluation configuration: {"Configuration":"Baselines (frame grouping)","Threshold":"τEER"}. This record does not assert independent reproduction.

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