Fix dimension order - PIL resize expects (width, height)

- PIL Image.resize() expects (width, height) order
- Model expects tensor shape [batch, height, width, channels] = [1, 448, 640, 3]
- Fixed resize to use (640, 448) for width=640, height=448
- Updated UI to clarify HxW format

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

app.py

@@ -44,8 +44,9 @@ def preprocess_image(image):
     """
     Preprocess image for Person Classification INT8 quantized model.
     """
-    # Resize to 448x640 as expected by the model
-    image = image.resize((448, 640))
+    # Resize to 640x448 (width x height) as PIL expects (width, height)
+    # Model expects input shape [batch, 448, 640, 3] meaning height=448, width=640
+    image = image.resize((640, 448))
     
     # Convert to numpy array
     img_array = np.array(image, dtype=np.float32)
@@ -380,7 +381,7 @@ with gr.Blocks(
                                 SRAM: ~1.5 MB<br>
                                 Flash: ~1.5 MB<br>
                                 Model: Person Classifier<br>
-                                Input: 448×640×3
+                                Input: 448×640×3 (HxW)
                             </div>
                         </div>
                     </div>