Image To Sketch

Summary

• Introduction
• Parameters
• Examples
• Usage for developers
• Reference
• License

Introduction

The task of Image-to-Sketch involves converting digital images into sketches, replicating the artistic style of hand-drawn or pencil sketches. This process is achieved through various algorithms and techniques in computer vision and image processing.

By converting images to sketches, this task provides a valuable tool for artists and designers to explore creative possibilities, experiment with different styles, and generate unique visual outputs. It allows for the fusion of traditional and digital art techniques, enabling the production of captivating and expressive sketches with the convenience and flexibility of digital media.

Parameters

Inputs

• input - (image -.png|.jpg|.jpeg): This is the original image that the user wants to convert into a sketch.
• ver - (text): This represents the specific drawing style that the user wants to apply to the transformation. It could be different predefined styles, such as "style 1" or "style 2".

Output

• output - (image -.png): This is the result of the image-to-sketch transformation based on the specified style.

Examples

input	ver	output
	style 1
	style 2

Usage for developers

Please find below the details to track the information and access the code for processing the model on our platform.

Requirements

python~=3.10
torch~=2.0.0

Code based on AIOZ structure

import torchvision.transforms as transforms
from PIL import Image
from .model import Generator
import torch, os

...
def do_ai_task(
        input: Union[str, Path],
        ver: Union[int, float],
        model_storage_directory: Union[str, Path],
        device: Literal["cpu", "cuda", "gpu"] = "cpu",
        *args, **kwargs) -> Any:
        
        model_id1 = os.path.abspath(model_storage_directory + "...")
        model1 = Generator(3, 1, 3)
        model1.load_state_dict(torch.load(model_id1), map_location=device))
        model1.eval()
        
        model_id2 = os.path.abspath(model_storage_directory + "...")
        model2 = Generator(3, 1, 3)
        model2.load_state_dict(torch.load(model_id2), map_location=device))
        model2.eval()
        
        input_img = Image.open(input).convert('RGB')
        transform = transforms.Compose([transforms.ToTensor()])
        input_img = transform(input_img)
        input_img = torch.unsqueeze(input_img, 0)

        drawing = 0
        with torch.no_grad():
            if ver == 'style 2':
                drawing = model2(input_img)[0].detach()
            else:
                drawing = model1(input_img)[0].detach()

        drawing = transforms.ToPILImage()(drawing)
        drawing.save("output.png")

        output = open("output.png", "rb")  # io.BufferedReader
    return output

Reference

This repository is based on and inspired by Caroline Chan's work. We sincerely thank them for sharing the code.

License

We respect and comply with the terms of the author's license cited in the Reference section.