Tools¶

visualize_gqcnn_dataset.py¶

Visualizes point clouds from a generated GQ-CNN training dataset, optionally filtering by the grasp robustness metrics.

Jeff Mahler

visualize_gqcnn_dataset.visualize_tensor_dataset(dataset, config)¶

Visualizes a Tensor dataset.

Parameters:	dataset (`TensorDataset`) – dataset to visualize config (`autolab_core.YamlConfig`) – parameters for visualization

Notes

Required parameters of config are specified in Other Parameters

Other Parameters:

field_name (str) – name of the field in the TensorDataset to visualize (defaults to depth_ims_tf_table, which is a single view point cloud of the object on a table)
field_type (str) – type of image that the field name correspondes to (defaults to depth, can also be segmask if using the field object_masks)
print_fields (list of str) – names of additiona fields to print to the command line
filter (dict mapping str to dict) – contraints that all displayed datapoints must satisfy (supports any univariate field name as a key and numeric thresholds)
gripper_width_px (float) – width of the gripper to plot in pixels
font_size (int) – size of font on the rendered images