KeypointNet: A Large-Scale 3D Keypoint Dataset (CVPR 2020)

KeypointNet is a large-scale and diverse 3D keypoint dataset that contains 8,234 3D models and 103K+ keypoints (103,447 in this release) from 16 object categories, aggregated from numerous human annotations on ShapeNet models.

• Paper: KeypointNet: A Large-Scale 3D Keypoint Dataset Aggregated From Numerous Human Annotations (CVPR 2020)
• arXiv: arXiv:2002.12687 · Hugging Face paper page
• Code and benchmarks: https://github.com/qq456cvb/KeypointNet
• Project page: https://qq456cvb.github.io/projects/keypointnet
• Interactive explorer: https://qq456cvb.github.io/projects/keypointnet#explore

Contents

• annotations/ — keypoint annotations as one JSON file per category, plus all.json with every category combined.
• pcds/ — sampled colored point clouds (2,048 points) for each ShapeNet model, one .pcd file per model at pcds/<class_id>/<model_id>.pcd.
• ShapeNetCore.v2.ply/ — colored triangle meshes (.ply with diffuse-texture vertex colors) for the annotated ShapeNet models, one file per model at ShapeNetCore.v2.ply/<class_id>/<model_id>.ply.
• knife_misaligned.txt — knives that are misaligned (x-axis flipped) in the original ShapeNet.

Covered categories: airplane (1022 models), bathtub (492), bed (146), bottle (380), cap (38), car (1002), chair (999), guitar (697), helmet (90), knife (270), laptop (439), motorcycle (298), mug (186), skateboard (141), table (1124) and vessel (910).

Annotation Format

Each category JSON is a list of annotated models:

[
    {
        "class_id": "03001627",  // WordNet id
        "model_id": "88382b877be91b2a572f8e1c1caad99e",
        "keypoints": [
            {
                "xyz": [0.16, 0.1, 0.1],   // keypoint coordinate
                "rgb": [255, 255, 255],    // keypoint color, uint8
                "semantic_id": 0,          // id of semantic meaning (consistent within a category)
                "pcd_info": { "point_index": 0 },          // index on the corresponding point cloud
                "mesh_info": { "face_index": 0, "face_uv": [0.2, 0.4, 0.4] }  // barycentric coords on the mesh face
            }
        ],
        "symmetries": {
            "reflection": [ { "kp_indexes": [0, 1] } ],
            "rotation": [
                {
                    "kp_indexes": [0, 1, 2, 3],
                    "is_circle": true,
                    "circle": { "center": [0.2, 0.5, 0.2], "radius": 0.32, "normal": [0, 1.0, 0] }
                }
            ]
        }
    }
]

Train/val/test splits and visualization/benchmark code are available in the GitHub repository.

Citation

@inproceedings{you2020keypointnet,
  title={KeypointNet: A Large-Scale 3D Keypoint Dataset Aggregated From Numerous Human Annotations},
  author={You, Yang and Lou, Yujing and Li, Chengkun and Cheng, Zhoujun and Li, Liangwei and Ma, Lizhuang and Lu, Cewu and Wang, Weiming},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  pages={13647--13656},
  year={2020}
}