Add inference endpoint

README.md

@@ -0,0 +1 @@
+Please use the image nvcr.io/nvidia/pytorch:21.11-py3 when you want to launch it

example_input.json

@@ -0,0 +1,3 @@
+{
+  "inputs": "https://media.lesechos.com/api/v1/images/view/5dadc9de8fe56f4ddf251469/1280x720/0602095339508-web-tete.jpg"
+}

handler.py

@@ -0,0 +1,53 @@
+from typing import Dict, List, Any
+from torchvision.models import resnet18, ResNet18_Weights
+from torchvision.io import read_image
+from PIL import Image
+import io
+import requests
+import torchvision.transforms.functional as transform
+
+from torch2trt import torch2trt
+from torchvision.models.alexnet import alexnet
+import torch
+
+# create some regular pytorch model...
+model = alexnet(pretrained=True).eval().cuda()
+
+# create example data
+x = torch.ones((1, 3, 224, 224)).cuda()
+
+# convert to TensorRT feeding sample data as input
+model_trt = torch2trt(model, [x])
+
+class EndpointHandler():
+    def __init__(self, path=""):
+        weights = ResNet18_Weights.DEFAULT
+        # create some regular pytorch model...
+        model = resnet18(weights=weights).eval().cuda()
+
+        # create example data
+        x = torch.ones((1, 3, 224, 224)).cuda()
+
+        # convert to TensorRT feeding sample data as input
+        self.pipeline = torch2trt(model, [x])
+        self.preprocess = weights.transforms()
+
+    def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
+        """
+       data args:
+            inputs (:obj: `str`)
+      Return:
+            A :obj:`list` | `dict`: will be serialized and returned
+        """
+        # get inputs
+        inputs = data.pop("inputs",data)
+        if inputs.startswith("http") or inputs.startswith("www"):
+            response = requests.get(inputs).content
+            img = transform.to_tensor(Image.open(io.BytesIO(response)))
+        else:
+            img = read_image(inputs)
+
+        batch = self.preprocess(img).unsqueeze(0)
+        prediction = self.pipeline(batch).squeeze(0).softmax(0)
+        
+        return prediction.tolist()