bop_sampling

Benchmark for 6D Object Pose Estimation (BOP)
Sampling objects, cameras and lights

This example shows how to load BOP objects and alternatingly sample light poses, camera poses (looking towards the objects) and object poses (including collision checks).

Usage

First make sure that you have downloaded a BOP dataset in the original folder structure. Also please clone the BOP toolkit.

In examples/bop_sampling/config.yaml add the path to your bop_toolkit clone to sys_paths and set the blender_install_path where Blender should be installed.

Execute in the BlenderProc main directory:

python run.py examples/bop_sampling/config.yaml <path_to_bop_data> examples/bop_sampling/output

• examples/bop_sampling/config.yaml: path to the pipeline configuration file.
• <path_to_bop_data>: path to a BOP dataset
• examples/bop_sampling/output: path to the output directory.

Visualization

Visualize the generated data and labels:

python scripts/visHdf5Files.py examples/bop_sampling/output/0.hdf5

Steps

• Loads object models and camera intrinsics from specified BOP dataset: loader.BopLoader module.
• Creates a point light sampled inside a shell: lighting.LightSampler module.
• Loops over: composite.CameraObjectSampler module.
  • Sample Object Poses inside a cube with collision checks
  • Sample Camera Poses inside a shell looking at objects
• Renders rgb: renderer.RgbRenderer module.
• Renders instance segmentation masks: renderer.SegMapRenderer module.
• Writes instance segmentation masks: writer.CocoAnnotationsWriter module.
• Writes labels and images to compressed hdf5 files in output_dir: writer.Hdf5Writer module.

Config file

BopLoader

If scene_id is not specified (default = -1), loader.BopLoader simply loads all or the specified obj_ids from the BOP dataset given by bop_dataset_path.

    {
      "name": "loader.BopLoader",
      "config": {
        "bop_dataset_path": "<args:0>",
        "mm2m": True,
        "split": "val",
        "obj_ids": [3,4,4,8],
        "model_type": ""
      }
    },

Following modules handle the sampling.

composite.CameraObjectSampler alternates between sampling new cameras using a camera.CameraSampler and sampling new object poses using a object.ObjectPoseSampler. Additionally, here you set the parameters

• noof_cams_per_scene after which the object poses are resampled
• total_noof_cams to generate

Alternatively, since we generated COCO annotations, you can also visualize the generated coco_annotations.json file:

python scripts/vis_coco_annotation.py /path/to/output_dir

More examples

• bop_scene_replication: Replicate the scenes and cameras from BOP datasets in simulation.