jyfgyjsgf

.gitignore

@@ -0,0 +1,7 @@
+.ipynb_checkpoints
+.vscode
+data
+data/training
+data/testing
+__pycache__
+nb.ipynb

app.py

@@ -1,7 +1,42 @@
 import gradio as gr
 
-def greet(name):
-    return "Hello " + name + "!!"
+import torch
+from torch.nn import Softmax
+from torchvision.io import read_image
+from torchvision.transforms.functional import rgb_to_grayscale
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
+from model import NeuralNet
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+model = NeuralNet().to(device)
+model.load_state_dict(torch.load("models/model.pth"))
+
+
+labels = [
+    "zero",
+    "one",
+    "two",
+    "three",
+    "four",
+    "five",
+    "six",
+    "seven",
+    "eight",
+    "nine"
+]
+
+def predict(image):
+    img = read_image(image).to(device)
+    img = rgb_to_grayscale(img) / 255
+    pred = model(img)[0]
+    prob = Softmax(dim=0)(pred).tolist()
+    return {l:p for l, p in zip(labels, prob)}
+
+
+iface = gr.Interface(
+    fn=predict,
+    inputs=gr.Image(label="Upload", type="filepath", shape=(28, 28)),
+    outputs=gr.Label(num_top_classes=10),
+    )
 iface.launch()

model.py

@@ -0,0 +1,18 @@
+import torch.nn as nn
+
+class NeuralNet(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.flatten = nn.Flatten()
+        self.layers = nn.Sequential(
+            nn.Linear(28*28, 512),
+            nn.ReLU(),
+            nn.Linear(512, 512),
+            nn.ReLU(),
+            nn.Linear(512, 10)
+        )
+
+    def forward(self, x):
+        x = self.flatten(x)
+        x = self.layers(x)
+        return x

models/model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d2226400fd63af4ccf83ed5c9f937cbe87ab5775bd42199f5270dc8da14acb
+size 2680775

requirements.txt

@@ -0,0 +1 @@
+torch