compressai_vision.compressai_vision.pipelines.fo_vcm.fo#

fiftyone#

compressai_vision.pipelines.fo_vcm.fo.predict.annexPredictions(predictors: list | None = None, fo_dataset: Dataset | None = None, gt_field: str = 'detections', predictor_fields: list | None = None, encoder_decoder=None, use_pb: bool = False, use_print: int = 1)[source]#

Run detector and EncoderDecoder instance on a dataset. Append detector results and bits-per-pixel to each sample.

Parameters:

  • predictors – A list of Detectron2 predictor. It can be a single element list for a single task or multiple elements for multi-task scenario

  • fo_dataset – Fiftyone dataset

  • gt_field – Which dataset member to use for ground truths. Default: “detections”

  • predictor_fields – Which dataset member to use for saving the Detectron2 results. Default: “detectron-predictions”. It also can be a list when evaluating multiple vision tasks.

  • encoder_decoder – (optional) a compressai_vision.evaluation.pipeline.EncoderDecoder subclass instance to apply on the image before detection

  • use_pb – Show progressbar or not. Nice for interactive runs, not so much for batch jobs. Default: False.

  • use_print – Print progress at every n:th. step. Default: 0 = no printing.

compressai_vision.pipelines.fo_vcm.fo.predict.annexVideoPredictions(predictors: list | None = None, fo_dataset: Dataset | None = None, gt_field: str = 'detections', predictor_fields: list | None = None, encoder_decoder=None, use_pb: bool = False, use_print: int = 1)[source]#

Run detector and EncoderDecoder instance on a dataset. Append detector results and bits-per-pixel to each sample.

Dataset.Sample.Frames

Parameters:

  • predictors – A list of Detectron2 predictor. It can be a single element list for a single task or multiple elements for multi-task scenario

  • fo_dataset – A fiftyone video dataset

  • gt_field – Which dataset member to use for ground truths. Default: “detections”

  • predictor_fields – Which dataset member to use for saving the Detectron2 results. Default: “detectron-predictions”. It also can be a list when evaluating multiple vision tasks.

  • encoder_decoder – (optional) a compressai_vision.evaluation.pipeline.EncoderDecoder subclass instance to apply on the image before detection

  • use_pb – Show progressbar or not. Nice for interactive runs, not so much for batch jobs. Default: False.

  • use_print – Print progress at every n:th. step. Default: 0 = no printing.

Video datasets look like this:

Name:        sfu-hw-objects-v1
Media type:  video
Num samples: 1
Persistent:  True
Tags:        []
Sample fields:
    id:         fiftyone.core.fields.ObjectIdField
    filepath:   fiftyone.core.fields.StringField
    tags:       fiftyone.core.fields.ListField(fiftyone.core.fields.StringField)
    metadata:   fiftyone.core.fields.EmbeddedDocumentField(fiftyone.core.metadata.VideoMetadata)
    media_type: fiftyone.core.fields.StringField
    class_tag:  fiftyone.core.fields.StringField
    name_tag:   fiftyone.core.fields.StringField
Frame fields:
    id:           fiftyone.core.fields.ObjectIdField
    frame_number: fiftyone.core.fields.FrameNumberField
    detections:   fiftyone.core.fields.EmbeddedDocumentField(fiftyone.core.labels.Detections)

Difference between image & video datasets

Image dataset:

Dataset
    first()
    __index__()
    iterator
    --> return Sample objects
        fields: id, filepath, ground-truths, detections, etc.

Video dataset

Dataset
    first()
    __index__
    iterator
    --> return Sample objects
        fields: id, filepath
        frames: Frames object
        --> __index__
            iterator
            --> returns Frame object
            fields: id, ground-truths, detections, etc.