nathankimnguyen412/houdini-lego-sdg

--- license: cc-by-4.0 task_categories: - object-detection - image-segmentation language: - en tags: - synthetic-data - lego - houdini - domain-randomization - coco - robotics pretty_name: Houdini Lego Synthetic Data Generation Dataset size_categories: - 10K<n<100K --- # Houdini Lego SDG Dataset Synthetic datasets generated by a procedural Houdini-based data generation pipeline for Lego brick instance segmentation and object detection. All images were rendered using Karma XPU in Houdini Solaris. Annotations were extracted automatically using Cryptomatte AOVs with no manual labeling on training data. **GitHub:** [nathankimnguyen412/houdini-sdg-pipeline](https://github.com/nathankimnguyen412/houdini-sdg-pipeline) ## Dataset Versions | Version | Images | Classes | Description | |:---------:|:------:|:-------:|-------------| | v1 | 1,000 | 1 | Baseline: no distractor objects, 29 domain randomization axes | | v2 | 1,000 | 1 | Distractor objects across 3 domain randomization axes added (bringing total axes from 29 to 32), 13 additional textures | | v3 | 3,000 | 7 | 7-class scale-up | | v4 | 7,000 | 7 | Full-scale primary dataset | ## Classes (v3 and v4) | ID | Name | |----|------| | 1 | 1x1_plate | | 2 | 2x4_plate | | 3 | 1x3_plate | | 4 | 1x2_plate_with_2_U_Clips | | 5 | 2x2_brick | | 6 | 1x1_brick_stud_on_side | | 7 | 1x1_round_brick | ## Format Each version contains an `images/` folder of 512×512 PNG renders and an `annotations.json` in COCO instance segmentation format. Each version also includes a `logs/wedge_log.csv` containing every domain randomization parameter value for every rendered scene, generated automatically by the TOPs/PDG pipeline during dataset generation. ## Citation See the GitHub repository for citation information. ## License CC BY 4.0 — free to use for any purpose with attribution.