api 기능 추가

README_anime.md

@@ -0,0 +1,4 @@
+# REAL-ESRGAN-API
+
+## Summary
+- Used ``inference_anime.py`` you can do easy upscaling anime img.

inference_anime.py

@@ -0,0 +1,137 @@
+import cv2
+from typing import Union
+import os
+from PIL import Image
+import numpy as np
+from basicsr.archs.rrdbnet_arch import RRDBNet
+from basicsr.utils.download_util import load_file_from_url
+from gfpgan import GFPGANer
+
+from realesrgan import RealESRGANer
+from realesrgan.archs.srvgg_arch import SRVGGNetCompact
+
+class SR_Webtoon:
+    def __init__(self,
+                 model_name_or_path:str="RealESRGAN_x4plus_anime_6B",
+                 outscale:int=4,
+                 suffix:str='out',
+                 tile:int=0,
+                 tile_pad:int=10,
+                 pre_pad:int=0,
+                 face_enhance:bool=False,
+                 fp32:bool=False,
+                 alpha_upsampler:bool='realesrgan',
+                 ext:str='auto',
+                 gpu_id:Union[str,None]=None):
+        """
+        Parameters
+        ----------
+        model_name_or_path : str
+            Model names: RealESRGAN_x4plus_anime_6B
+        outscale : float
+            The final upsampling scale of the image
+        suffix : str
+            Suffix of the restored image
+        tile : int
+            Tile size, 0 for no tile during testing
+        tile_pad : int
+            Tile padding
+        pre_pad : int
+            Pre padding size at each border
+        face_enhance : bool
+            Use GFPGAN to enhance face
+        fp32 : bool
+            Use fp32 precision during inference. Default: fp16 (half precision).
+        alpha_upsampler : str
+            The upsampler for the alpha channels. Options: realesrgan | bicubic
+        ext : str
+            Image extension. Options: auto | jpg | png, auto means using the same extension as inputs
+        gpu_id : int
+            gpu device to use (default=None) can be 0,1,2 for multi-gpu
+        """
+        self.__model_name_or_path = model_name_or_path
+        self.__outscale = outscale
+        self.__suffix = suffix
+        self.__tile = tile
+        self.__tile_pad = tile_pad
+        self.__pre_pad = pre_pad
+        self.__face_enhance = face_enhance
+        self.__fp32 = fp32
+        self.__alpha_upsampler = alpha_upsampler
+        self.__ext = ext
+        self.__gpu_id = gpu_id
+
+        # model select
+        self.__upsampler = self.__model_select()
+
+        # face
+        if self.__face_enhance:
+            self.__upsampler = self.__face_enhancer()
+
+
+    def __model_select(self)->RealESRGANer:
+        if self.__model_name_or_path == "RealESRGAN_x4plus_anime_6B":
+            # download
+            file_url = ['https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth']
+
+            # model file format
+            model_path = os.path.join('weights', self.__model_name_or_path+ '.pth')
+            if not os.path.isfile(model_path):
+                ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
+                for url in file_url:
+                    # model_path will be updated
+                    model_path = load_file_from_url(
+                        url=url, model_dir=os.path.join(ROOT_DIR, 'weights'), progress=True, file_name=None)
+
+        elif self.__model_name_or_path is not None:
+            model_path = self.__model_name_or_path
+        else:
+            raise ValueError("model_name_or_path must be RealESRGAN_x4plus_anime_6B")
+
+        # RealESRGAN_x4plus_anime_6B
+        model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=6, num_grow_ch=32, scale=4)
+        netscale = 4
+
+        # restorer
+        upsampler = RealESRGANer(
+            scale=netscale,
+            model_path=model_path,
+            dni_weight=None,
+            model=model,
+            tile=self.__tile,
+            tile_pad=self.__tile_pad,
+            pre_pad=self.__pre_pad,
+            half=not self.__fp32,
+            gpu_id=self.__gpu_id)
+        return upsampler
+
+    def __face_enhancer(self)->GFPGANer:
+        face_enhancer = GFPGANer(
+            model_path='https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.3.pth',
+            upscale=self.__outscale,
+            arch='clean',
+            channel_multiplier=2,
+            bg_upsampler=self.__upsampler)
+        return face_enhancer
+
+    def sr(self,image:Image):
+        # image processing
+        image = image.convert("RGBA")
+        img = np.array(image)
+
+        if self.__face_enhance:
+            _, _, output = self.__upsampler.enhance(img, has_aligned=False, only_center_face=False, paste_back=True)
+        else:
+            output, _ = self.__upsampler.enhance(img, outscale=self.__outscale)
+
+        sr_img = Image.fromarray(output)
+        return sr_img
+
+# test
+
+# if __name__ == "__main__":
+#     SR = SR_Webtoon()
+#     img = Image.open("/content/004_nocap.jpg").resize((512,512))
+#     print(img)
+#     r = SR.sr(img)
+#     r.save("./test1.png")