# Scene Graph Generation on PVSG IoU threshold = 50 (test) Source: [Scene Graph Generation on PVSG IoU threshold = 50 (test)]() Collection index: [llms.txt]() Data updated: 2026-05-14T09:46:24.556Z Evaluation of Scene Graph Generation performance on the PVSG test split, specifically using an IoU threshold of 50 for evaluation. Metrics include detailed object and relationship precision, recall, and F1 scores (strict and soft), alongside generation time and failure rate. 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 | Scene Graph Generation | | Dataset (including any stated split) | PVSG IoU threshold = 50 (test) | | Metric | Generation Time (s) | | Unit | TIME | | Direction | Lower is better | | Results loaded | 6 of 6 | | 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 - [Generation Time (s)](): Lower is better; unit: TIME. - [Failure Rate (%)](): Lower is better; unit: PERCENT. - [Object Precision (Strict)](): Higher is better; unit: PERCENT. - [Object Precision (Soft)](): Higher is better; unit: PERCENT. - [Object Recall (Strict)](): Higher is better; unit: PERCENT. - [Object Recall (Soft)](): Higher is better; unit: PERCENT. - [Object F1 Score (Strict)](): Higher is better; unit: PERCENT. - [Object F1 Score (Soft)](): Higher is better; unit: PERCENT. - [Relationship Precision (Strict)](): Higher is better; unit: PERCENT. - [Relationship Precision (Soft)](): Higher is better; unit: PERCENT. - [Relationship Recall (Strict)](): Higher is better; unit: PERCENT. - [Relationship Recall (Soft)](): Higher is better; unit: PERCENT. - [Relationship F1 Score (Strict)](): Higher is better; unit: PERCENT. - [Relationship F1 Score (Soft)](): Higher is better; unit: PERCENT. - [SGG Score (Strict)](): Higher is better; unit: PERCENT. - [SGG Score (Soft)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Generation Time (s) | Result date | Report paper | | --- | --- | --- | --- | | SceneGraphVLM (SFT) | 1.068 | Not recorded | 6a055729dbc836b87d828da5 | | SceneGraphVLM (SFT+RL) | 1.068 | Not recorded | 6a055729dbc836b87d828da5 | | SceneGraphVLM (SFT) | 1.068 | Not recorded | 6a055729dbc836b87d828da5 | | SceneGraphVLM (SFT+RL) | 1.068 | Not recorded | 6a055729dbc836b87d828da5 | | GPT-5.4-mini | 2.4 | Not recorded | 6a055729dbc836b87d828da5 | | GPT-5.4-nano | 2.678 | Not recorded | 6a055729dbc836b87d828da5 | ### SceneGraphVLM (SFT) Generation Time (s): 1.068. Result ID: 6a0599ee47f34ce924ed0c5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompting":"GT prompt","Train split":"PSFR"}. This record does not assert independent reproduction. ### SceneGraphVLM (SFT+RL) Generation Time (s): 1.068. Result ID: 6a0599ee47f34ce924ed0c5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompting":"GT prompt","Train split":"BaseAnnot"}. This record does not assert independent reproduction. ### SceneGraphVLM (SFT) Generation Time (s): 1.068. Result ID: 6a0599ee47f34ce924ed0c5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompting":"Generated prompt","Train split":"BaseAnnot"}. This record does not assert independent reproduction. ### SceneGraphVLM (SFT+RL) Generation Time (s): 1.068. Result ID: 6a0599ee47f34ce924ed0c60. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompting":"Generated prompt","Train split":"BaseAnnot"}. This record does not assert independent reproduction. ### GPT-5.4-mini Generation Time (s): 2.4. Result ID: 6a0599ee47f34ce924ed0c5e. [Reporting paper]() Recorded evaluation configuration: {"Prompting":"Generated prompt"}. This record does not assert independent reproduction. ### GPT-5.4-nano Generation Time (s): 2.678. Result ID: 6a0599ee47f34ce924ed0c5b. [Reporting paper]() Recorded evaluation configuration: {"Prompting":"GT prompt"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.