:tada: first commit

README.md

@@ -1,3 +1,34 @@
+# ---
+language:
+- ru
+- en
+tags:
+- PyTorch
+thumbnail: "https://github.com/sberbank-ai/Real-ESRGAN"
 ---
-license: openrail
----
+# Real-ESRGAN
+
+PyTorch implementation of a Real-ESRGAN model trained on custom dataset. This model shows better results on faces compared to the original version. It is also easier to integrate this model into your projects.
+
+Real-ESRGAN is an upgraded ESRGAN trained with pure synthetic data is capable of enhancing details while removing annoying artifacts for common real-world images.
+
+- [Paper](https://arxiv.org/abs/2107.10833)
+- [Original implementation](https://github.com/xinntao/Real-ESRGAN)
+- [Our github](https://github.com/LikeRainDay/Real-ESRGAN)
+
+## Usage
+
+Code for using model you can obtain in our [repo](https://github.com/LikeRainDay/Real-ESRGAN).
+```python
+import torch
+from PIL import Image
+import numpy as np
+from RealESRGAN import RealESRGAN
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+model = RealESRGAN(device, scale=4)
+model.load_weights('weights/RealESRGAN_x4.pth', download=True)
+path_to_image = 'inputs/lr_image.png'
+image = Image.open(path_to_image).convert('RGB')
+sr_image = model.predict(image)
+sr_image.save('results/sr_image.png')
+```

RealESRGAN_x2plus.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49fafd45f8fd7aa8d31ab2a22d14d91b536c34494a5cfe31eb5d89c2fa266abb
+size 67061725

RealESRGAN_x4plus.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fa0d38905f75ac06eb49a7951b426670021be3018265fd191d2125df9d682f1
+size 67040989

RealESRGAN_x8.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b72fb469d12f05a4770813d2603eb1b550f40df6fb8b37d6c7bc2db3d2bff5e
+size 67189359

handler.py

@@ -0,0 +1,51 @@
+from base64 import b64encode, b64decode
+from io import BytesIO
+from pathlib import Path
+
+import numpy as np
+from basicsr.archs.rrdbnet_arch import RRDBNet
+from PIL import Image
+from realesrgan import RealESRGANer
+
+
+class EndpointHandler:
+    def __init__(self, path=""):
+        model = RRDBNet(num_in_ch=3, num_out_ch=3)
+        self.upsampler = RealESRGANer(
+            scale=4,
+            model_path=str(Path(path) / "RealESRGAN_x4plus.pth"),
+            model=model,
+            tile=0,
+            pre_pad=0,
+            half=True,
+        )
+
+    def __call__(self, data):
+        """
+        Args:
+            data (:obj:):
+                includes the input data and the parameters for the inference.
+        Return:
+            A :obj:`dict`:. base64 encoded image
+        """
+        image = data.pop("inputs", data)
+
+        # This lets us pass local images as well while developing
+        if isinstance(image, str):
+            image = Image.open(BytesIO(b64decode(image)))
+        elif isinstance(image, bytes):
+            image = Image.open(BytesIO(image))
+
+        image = np.array(image)
+        image = image[:, :, ::-1]  # RGB -> BGR
+        image, _ = self.upsampler.enhance(image, outscale=4)
+        image = image[:, :, ::-1]  # BGR -> RGB
+        image = Image.fromarray(image)
+
+        # Turn output image into bytestr
+        buffered = BytesIO()
+        image.save(buffered, format="PNG")
+        img_bytes = b64encode(buffered.getvalue())
+        img_str = img_bytes.decode()
+
+        return {"image": img_str}

requirements.txt

@@ -0,0 +1 @@
+realesrgan==0.3.0