Image processing pipeline

Modular image processing pipeline using OpenCV and Python generators. Inspired by Albumentations framework to process images

Setup environment

This project is using virtualenv for project environment management.

Setup the project environment:

$ virtualenv -p python3.8 venv
$ source venv/bin/activate

Getting started

For stylization pipeline:

Create config.yaml with content:

_target_: pipeline.compose.Compose
nodes:
  - _target_: pipeline.readers.opencv_read_image.ReadOpenCVImage
  - _target_: pipeline.preprocess.resize2division.Resize2Dividable
    must_divided: &must_divided 32
  - _target_: pipeline.detectors.face_cropping.FaceCropping
    detector:
      _target_: pipeline.detectors.lib.mediapipe_detector.StatMediaPipeDetector
      target_size: &target_shape 256
      must_divide: *must_divided
  - _target_: pipeline.stylization.image_gan_stylization.ImageGANStylization
    inference_engine: &inference_engine
      _target_: pipeline.stylization.inference_engine.inference_engine_cache.get_onnx_inference
      model_path: # path to your onnx model
  - _target_: pipeline.stylization.bbox_gan_stylization.BBoxGANStylization
    inference_engine: *inference_engine
  - _target_: pipeline.postprocess.merging_crops.seamless_merging_crops.SeamlessMergingCrops
    crops_size: *target_shape

In your python script use next code


import logging
import os
from pathlib import Path

import cv2
import numpy as np
from omegaconf import OmegaConf

from pipeline.pipeline import Pipeline
from pipeline.processed_data import ImagePipelineData
from pipeline.utils.pipeline_data import get_data
from tests.pipeline import CONFIG
from utils.instantiate import instantiate


config_path = Path(datadir) / 'config.yaml'

config = OmegaConf.load(config_path)
pipeline: Pipeline = instantiate(config)


data = get_data(im_path)
result: ImagePipelineData = pipeline(data)
result_image = result.processed_image

cv2.imwrite(os.path.join(out_dir, img_name), cv2.cvtColor(result.processed_image, cv2.COLOR_RGB2BGR))

Tests

pytest is used as a test framework. All tests are stored in tests folder. Run the tests:

$ pytest

Resources and Credits

For Unix like pipeline idea credits goes to this Gist
Source of the example images and videos is pixbay
Some ideas and code snippets are borrowed from pyimagesearch
Color constants from Python Color Constants Module

License