feat: Enhance README and app functionality; add new images and update requirements

- Enhance app.py for zerogpu ssupport
- Enhance README with image lincence
- Chore requirements.txt
- Add cat, hot-dog, llama, medieval_knight images

README.md

@@ -10,4 +10,14 @@ pinned: false
 license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference
+
+contains free-licensed images, downloaded from [unsplash](https://unsplash.com/). Curated and created by:
+
+[cat](https://unsplash.com/photos/black-and-white-cat-lying-on-brown-bamboo-chair-inside-room-gKXKBY-C-Dk)
+
+[hot-dog](https://unsplash.com/photos/hotdog-sandwich-on-white-ceramic-plate-w96PYF0Uwjs)
+
+[medieval_knight](https://huggingface.co/datasets/thliang01/medieval_knight)
+
+[llama](https://huggingface.co/datasets/thliang01/Cute-Llama)

app.py

@@ -1,7 +1,6 @@
+import spaces
 import torch
-
-model = torch.hub.load('pytorch/vision:v0.6.0', 'resnet18', pretrained=True).eval()
-
+import gradio as gr
 import requests
 from torchvision import transforms
 
@@ -9,17 +8,32 @@ from torchvision import transforms
 response = requests.get("https://git.io/JJkYN")
 labels = response.text.split("\n")
 
-def predict(inp):
-  inp = transforms.ToTensor()(inp).unsqueeze(0)
-  with torch.no_grad():
-    prediction = torch.nn.functional.softmax(model(inp)[0], dim=0)
-    confidences = {labels[i]: float(prediction[i]) for i in range(1000)}    
-  return confidences
+# Load model initially (will be moved to GPU dynamically)
+model = torch.hub.load('pytorch/vision:v0.6.0', 'resnet34', pretrained=True).eval()
 
-import gradio as gr
+@spaces.GPU
+def predict(inp):
+    """
+    This function will be allocated GPU resources dynamically when called
+    """
+    # Move model to GPU when function is called
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    model.to(device)
+    
+    # Process input image
+    inp = transforms.ToTensor()(inp).unsqueeze(0).to(device)
+    
+    with torch.no_grad():
+        prediction = torch.nn.functional.softmax(model(inp)[0], dim=0)
+        confidences = {labels[i]: float(prediction[i]) for i in range(1000)}
+    
+    return confidences
 
-gr.Interface(fn=predict, 
-             inputs=gr.Image(type="pil"),
-             outputs=gr.Label(num_top_classes=3),
-             examples=["lion.jpg", "cheetah.jpg"],
-             css=".footer{display:none !important}").launch()
+# Create Gradio interface
+gr.Interface(
+    fn=predict, 
+    inputs=gr.Image(type="pil"),
+    outputs=gr.Label(num_top_classes=3),
+    examples=["lion.jpg", "cheetah.jpg", "cat.avif", "hot-dog.avif", "llama.jpg", "medieval_knight.jpg"],
+    css=".footer{display:none !important}"
+).launch()

requirements.txt

@@ -1 +1,5 @@
-torch
+torch
+gradio
+spaces
+numpy
+torchvision