commiting the meme classification hf demo

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+scripts_and_models/efficientNet_clf.pt filter=lfs diff=lfs merge=lfs -text
+example_imgs/* filter=lfs diff=lfs merge=lfs -text
+example_imgs/*.jpg filter=lfs diff=lfs merge=lfs -text
+example_imgs/meme.png filter=lfs diff=lfs merge=lfs -text
+example_imgs/non-meme.jpg filter=lfs diff=lfs merge=lfs -text

scripts_and_models/app.py

@@ -0,0 +1,65 @@
+### 1. Imports and class names setup ### 
+import gradio as gr
+import os
+import torch
+from pathlib import Path 
+from timeit import default_timer as timer
+from typing import Tuple, Dict
+from torchvision import transforms
+
+class_names=['meme', 'non-meme']
+
+
+model_path=Path("efficientNet_clf.pt")
+model = torch.jit.load(model_path)
+image_transform = transforms.Compose([
+            transforms.Resize((224,224)),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                                 std=[0.229, 0.224, 0.225]),
+        ])
+print(image_transform)
+
+def predict(img) -> Tuple[Dict, float]:
+    """Transforms and performs a prediction on img and returns prediction and time taken.
+    """
+
+    print("---img path is: ",img)
+    start_time = timer()    
+    model.to("cpu")
+    model.eval()
+    with torch.inference_mode():
+        img = image_transform(img).unsqueeze(dim=0)
+        pred_probs = torch.softmax(model(img).to("cpu"), dim=1)
+    
+    pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}
+    pred_time = round(timer() - start_time, 5)
+    
+    return pred_labels_and_probs, pred_time
+    
+        #print(e)
+        #return "error",0
+        
+        
+        
+title = "Meme classifiication"
+description = "An EfficientNetB2 model to classify images of food into 2 classes:meme and non-meme"
+
+
+example_list = ["../example_imgs/"+i for i in os.listdir("../example_imgs")]
+#print(example_list)
+
+demo = gr.Interface(
+    fn=predict,
+    inputs=gr.Image(type="pil"),
+    outputs=[
+        gr.Label(num_top_classes=2, label="Predictions"),
+        gr.Number(label="Prediction time (s)"),
+    ],
+    examples=example_list,
+    title=title,
+    description=description,
+)
+
+demo.launch()
+#predict(example_list[0])

scripts_and_models/efficientNet_clf.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00aa3d1e2f5828f9529424a021577e181b779d77fc95a47ecc3d9f562d3b9b7e
+size 16535370

scripts_and_models/inference.py

@@ -0,0 +1,61 @@
+from typing import List, Tuple
+from PIL import Image
+import torch
+import torchvision
+from torchvision import datasets, transforms
+import matplotlib.pyplot as plt
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+
+
+def pred_and_plot_image(model: torch.nn.Module,
+                        image_path: str, 
+                        class_names: List[str],
+                        image_size: Tuple[int, int] = (224, 224),
+                        transform: torchvision.transforms = None,
+                        device: torch.device=device):
+    
+
+    img = Image.open(image_path)
+
+    if transform is not None:
+        image_transform = transform
+    else:
+        image_transform = transforms.Compose([
+            transforms.Resize(image_size),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                                 std=[0.229, 0.224, 0.225]),
+        ])
+
+
+    model.to(device)
+    model.eval()
+    with torch.inference_mode():
+      transformed_image = image_transform(img).unsqueeze(dim=0)
+      target_image_pred = model(transformed_image.to(device))
+    target_image_pred_probs = torch.softmax(target_image_pred, dim=1)
+    target_image_pred_label = torch.argmax(target_image_pred_probs, dim=1)
+
+    plt.figure()
+    plt.imshow(img)
+    plt.title(f"Pred: {class_names[target_image_pred_label]} | Prob: {target_image_pred_probs.max():.3f}")
+    plt.axis(False);
+    plt.show()
+    
+    
+    
+    
+from pathlib import Path    
+
+model_path=Path("efficientNet_clf.pt")
+print(model_path)
+
+model = torch.jit.load(model_path)
+
+class_names=['meme', 'non-meme']
+
+pred_and_plot_image(model=model,
+                    image_path="../example_imgs/meme.png",
+                    class_names=class_names)