rofunc.utils.visualab.segment.vlpart.vlpart#

1.  Module Contents#

1.1.  Classes#

VLPart

1.2.  Functions#

build_vlpart

custom_detector_postprocess

detector_postprocess with support on global_masks

1.3.  API#

rofunc.utils.visualab.segment.vlpart.vlpart.build_vlpart(checkpoint=None)[source]#
class rofunc.utils.visualab.segment.vlpart.vlpart.VLPart(backbone, proposal_generator: torch.nn.Module, roi_heads: torch.nn.Module)[source]#

Bases: torch.nn.Module

property device#
get_text_embeddings(vocabulary, prefix_prompt='a ')[source]#
inference(batched_inputs: List[Dict[str, torch.Tensor]], do_postprocess: bool = True, text_prompt: str = 'dog')[source]#
preprocess_image(batched_inputs: List[Dict[str, torch.Tensor]])[source]#

Normalize, pad and batch the input images.

rofunc.utils.visualab.segment.vlpart.vlpart.custom_detector_postprocess(results, output_height: int, output_width: int, max_shape, mask_threshold: float = 0.5)[source]#

detector_postprocess with support on global_masks