initial commit

.gitignore

@@ -0,0 +1,3 @@
+# PhpStorm / IDEA
+.idea
+

README.md

@@ -1,3 +1,38 @@
 ---
-license: apache-2.0
+tags:
+  - vision
+  - image-to-image
+  - endpoints-template
+inference: false
+pipeline_tag: image-to-image
+base_model: caidas/swin2SR-classical-sr-x2-64
+library_name: generic
 ---
+
+# Fork of [caidas/swin2SR-classical-sr-x2-64](https://huggingface.co/caidas/swin2SR-classical-sr-x2-64) for a `image-to-image` Inference endpoint.
+
+> Inspired by https://huggingface.co/sergeipetrov/swin2SR-classical-sr-x2-64-IE
+
+This repository implements a `custom` task for `image-to-image` for 🤗 Inference Endpoints to allow image up scaling by doubling image resolution. 
+The code for the customized pipeline is in the handler.py.
+
+To use deploy this model an Inference Endpoint you have to select `Custom` as task to use the `handler.py` file. 
+
+### expected Request payload
+
+Image to be labeled as binary. 
+
+#### CURL
+
+```
+curl URL \
+        -X POST \
+        --data-binary @car.png \
+        -H "Content-Type: image/png"
+```
+
+#### Python
+
+```python
+requests.post(ENDPOINT_URL, headers={"Content-Type": "image/png"}, data=open("car.png", 'rb').read()).json()
+```

handler.py

@@ -0,0 +1,48 @@
+from typing import Dict, List, Any
+from transformers import AutoImageProcessor, Swin2SRForImageSuperResolution
+import torch
+import base64
+import logging
+import numpy as np
+from PIL import Image
+from io import BytesIO
+
+logger = logging.getLogger()
+logger.setLevel(logging.DEBUG)
+
+# check for GPU
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+
+class EndpointHandler:
+    def __init__(self, path=""):
+        # load the model
+        self.processor = AutoImageProcessor.from_pretrained("caidas/swin2SR-classical-sr-x2-64")
+        self.model = Swin2SRForImageSuperResolution.from_pretrained("caidas/swin2SR-classical-sr-x2-64")
+        # move model to device
+        self.model.to(device)
+
+    def __call__(self, data: Any):
+        """
+        Args:
+            data (:obj:):
+                binary image data to be labeled
+        Return:
+            A :obj:`string`:. Base64 encoded image string
+        """
+
+        image = data["inputs"]
+        inputs = self.processor(image, return_tensors="pt").to(device)
+        with torch.no_grad():
+            outputs = self.model(**inputs)
+        
+        output = outputs.reconstruction.data.squeeze().float().cpu().clamp_(0, 1).numpy()
+        output = np.moveaxis(output, source=0, destination=-1)
+        output = (output * 255.0).round().astype(np.uint8)
+
+        img = Image.fromarray(output)
+        buffered = BytesIO()
+        img.save(buffered, format="JPEG")
+        img_str = base64.b64encode(buffered.getvalue())
+
+        return img_str.decode()