first commit

app.py

@@ -0,0 +1,52 @@
+import gradio as gr
+import os
+import torch
+
+from model import create_vit
+from timeit import default_timer as timer
+from typing import Tuple, Dict
+
+class_names = ["buildings", "forest", "glacier", "mountain", "sea", "street"]
+
+vit_model, vit_transforms = create_vit(num_classes=len(class_names),
+                                       seed=42)
+
+vit_model.load_state_dict(
+    torch.load(
+        f="pretrained_vit_feature_extractor_scene_recognition.pth", 
+        map_location=torch.device("cpu")
+    )
+)
+
+def predict(img):
+    start_timer = timer()
+    
+    img = vit_transforms(img).unsqueeze(0)
+    
+    vit_model.eval()
+    with torch.inference_mode():
+        pred_prob = torch.softmax(vit_model(img), dim=1)
+        
+    pred_labels_and_probs = {class_names[i]: float(pred_prob[0][i]) for i in range(len(class_names))}
+    
+    pred_time = round(timer() - start_timer, 5)
+    
+    return pred_labels_and_probs, pred_time
+
+title = "Scene Recognition: Intel Image Classification"
+description = "A ViT feature extractor Computer Vision model to classify images of scenes from 1 out of 6 classes."
+article = "Access project repository at [GitHub](https://github.com/Ammar2k/intel_image_classification)"
+
+example_list = [["examples/" + example] for example in os.listdir("examples")]
+
+demo = gr.Interface(fn=predict, 
+                    inputs=gr.Image(type="pil"),
+                    outputs=[gr.Label(num_top_classes=6, label="Predictions"), 
+                    gr.Number(label="Prediction time(s)")],
+                    examples=example_list,
+                    title=title,
+                    description=description,
+                    article=article
+                   )
+
+demo.launch()

model.py

@@ -0,0 +1,18 @@
+import torch
+import torchvision
+
+def create_vit(num_classes: int=6,
+               seed: int=42):
+    weights = torchvision.models.ViT_B_16_Weights.DEFAULT
+    
+    transforms = weights.transforms()
+    
+    model = torchvision.models.vit_b_16(weights=weights)
+    
+    for param in model.parameters():
+        param.requires_grad = False
+        
+    torch.manual_seed(seed)
+    model.heads = torch.nn.Sequential(torch.nn.LayerNorm(normalized_shape=768),
+                                      torch.nn.Linear(in_features=768, out_features=num_classes))
+    return model, transforms

pretrained_vit_feature_extractor_scene_recognition.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a388a90788d45eee524052ed07aede4d9c39a540951c87b269224d907b47005
+size 343286657

requirements.txt

@@ -0,0 +1,3 @@
+torch==2.0.0
+torchvision==0.15.0
+gradio==3.34.0