100 sample images from a larger 50,000-image synthetic dataset for 6-degree-of-freedom pose estimation. The dataset features the ABB dino robot model with 17 keypoints, rendered in Blender 5.1 EEVEE at 640×640 resolution. Annotations are provided in COCO keypoint format for the sample set.
Use Cases
- Train or fine-tune 6-DoF pose estimation models based on the synthetic robot images.
- Benchmark keypoint detection algorithms based on the 17 annotated keypoints for the ABB dino.
- Validate model performance on synthetic data before deployment on real-world robotic systems.
- Develop data augmentation pipelines using the high-resolution 640×640 synthetic renders.
Strengths
- Contains 100 sample images from a larger 50,000-image synthetic dataset.
- Images are rendered at a consistent 640×640 resolution using Blender 5.1 EEVEE.
- Annotations are provided in a standard COCO keypoint format for the samples.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- The sample size of 100 images may be insufficient for training a model from scratch.
Provenance
- Source
- huggingface
- Collection Method
- Synthetically generated using the abb_pose tool and Blender 5.1 EEVEE renderer.
- Time Range
- null
- Freshness
- Last updated 2026-04-01 02:00:57; freshness should be verified.
- Geography
- null