GTASA: Ground Truth Annotations for Spatiotemporal Analysis, Evaluation and Training of Video Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917401196494848 |
|---|---|
| author | Cudlenco, Nicolae Masala, Mihai Leordeanu, Marius |
| author_facet | Cudlenco, Nicolae Masala, Mihai Leordeanu, Marius |
| contents | Generating complex multi-actor scenario videos remains difficult even for state-of-the-art neural generators, while evaluating them is hard due to the lack of ground truth for physical plausibility and semantic faithfulness. We introduce GTASA, a corpus of multi-actor videos with per-frame spatial relation graphs and event-level temporal mappings, and the system that produced it based on Graphs of Events in Space and Time (GEST): GEST-Engine. We compare our method with both open and closed source neural generators and prove both qualitatively (human evaluation of physical validity and semantic alignment) and quantitatively (via training video captioning models) the clear advantages of our method. Probing four frozen video encoders across 11 spatiotemporal reasoning tasks enabled by GTASA's exact 3D ground truth reveals that self-supervised encoders encode spatial structure significantly better than VLM visual encoders. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_10385 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GTASA: Ground Truth Annotations for Spatiotemporal Analysis, Evaluation and Training of Video Models Cudlenco, Nicolae Masala, Mihai Leordeanu, Marius Computer Vision and Pattern Recognition Generating complex multi-actor scenario videos remains difficult even for state-of-the-art neural generators, while evaluating them is hard due to the lack of ground truth for physical plausibility and semantic faithfulness. We introduce GTASA, a corpus of multi-actor videos with per-frame spatial relation graphs and event-level temporal mappings, and the system that produced it based on Graphs of Events in Space and Time (GEST): GEST-Engine. We compare our method with both open and closed source neural generators and prove both qualitatively (human evaluation of physical validity and semantic alignment) and quantitatively (via training video captioning models) the clear advantages of our method. Probing four frozen video encoders across 11 spatiotemporal reasoning tasks enabled by GTASA's exact 3D ground truth reveals that self-supervised encoders encode spatial structure significantly better than VLM visual encoders. |
| title | GTASA: Ground Truth Annotations for Spatiotemporal Analysis, Evaluation and Training of Video Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.10385 |