Crate caffe2_image

source ·

Structs

  • BoundingBox
  • ImageInputOp
    | Imports and processes images from a | database. For each run of the operator, | batch_size images will be processed. | GPUs can optionally be used for part | of the processing. | | The following transformations are | applied to the image | | - A bounding box is applied to the initial | image (optional) | | - The image is rescaled either up or down | (with the scale argument) or just up | (with the minsize argument) | | - The image is randomly cropped (crop | size is passed as an argument but the | location of the crop is random except | if is_test is passed in which case the | image in cropped at the center) | | - The image is normalized. Each of its | color channels can have separate normalization | values | | The dimension of the output image will | always be cropxcrop |
  • PerImageArg
    | Structure to store per-image information | | This can be modified by the DecodeAnd* | so needs to be privatized per launch. |

Enums

  • ImageInputOpLabelType
    | SINGLE_LABEL: single integer label | for multi-class classification | | MULTI_LABEL_SPARSE: sparse active | label indices for multi-label classification | MULTI_LABEL_DENSE: dense label embedding | vector for label embedding regression | MULTI_LABEL_WEIGHTED_SPARSE: sparse | active label indices with per-label | weights for multi-label classification | | SINGLE_LABEL_WEIGHTED: single integer | label for multi-class classification | with weighted sampling EMBEDDING_LABEL: | an array of floating numbers representing | dense embedding. | | It is useful for model distillation |
  • ScaleJitterType
    | INCEPTION_STYLE: Random crop with | size 8% - 100% image area and aspect ratio | in [3/4, 4/3]. Reference: GoogleNet | paper |

Functions