gen ocr rec data

gen aug data

Randomly choose one of the best 24 Sub-policies on ImageNet. Example: >>> policy = ImageNetPolicy() >>> transformed = policy

Single level attribute dict, NOT recursive

AugMix Transform Adapted and improved from impl here: https://github.com/google-research/augmix/blob/master/imagenet.py From paper: 'AugMix:

Directed Acyclic Graph(DAG) engine, builds one DAG topology.

decode image

The executor which implements model series pipeline Args: env_cfg: The enrionment configuration model_cfg: The models config

class to hold information about each frame

extract key frames based on sum of absolute differences in LUV colorspace.

OperatorParamError

OperatorParamError

OperatorParamError

RandAugment wrapper to auto fit different img types

RandomErasing.

RandomErasing for OCR data.

resize image

Perform AugMix augmentation and compute mixture.

ImageNetPolicy wrapper to auto fit different img types

Base Operator, implement of prediction process Args

Convert between text-label and text-index

BatchOperator

BboxCropOp

Randomly choose one of the best 25 Sub-policies on CIFAR10. Example: >>> policy = CIFAR10Policy() >>> transformed = policy(i

Convert between text-label and text-index

rotate

ColorJitter.

crop image

crop image

crop image

crop image and padding to original size

Cutmix operator reference: https://arxiv.org/abs/1905.04899

The post process for Differentiable Binarization (DB).

decode image

decode image

FMix augmentation Args: decay_power (float): Decay power for frequency decay prop 1/f**d alpha (float): Alpha value

Fmix operator reference: https://arxiv.org/abs/2002.12047

FragmentCompositionOp

Builds a queue out of a data generator. Args: generator: a generator function which endlessly yields data use_multiprocessin

KeyFrameExtractionOp

Mixup operator reference: https://arxiv.org/abs/1710.09412

Base Operator, implement of prediction process Args