| task_categories: | |
| - visual-question-answering | |
| - object-detection | |
| language: | |
| - en | |
| tags: | |
| - LEGO | |
| - Multitask | |
| pretty_name: LEGO_Reader | |
| size_categories: | |
| - 1K<n<10K | |
| Completed Dataset Version 1, aimed at boosting model performance in 3 key areas: | |
| Object Detection: 24k instances | |
| Grounding (Scene Understanding): 8k instances | |
| State Detection: 12k instances | |