Update app.py

app.py

@@ -1,82 +1,33 @@
 import cv2
-import numpy as np
 import gradio as gr
 
-# Global variables to track the state of manual segmentation
-drawing = False
-roi_points = []
-
-# Function to handle mouse events
-def mouse_event(event, x, y, flags, param):
-    global roi_points, drawing
-    if event == cv2.EVENT_LBUTTONDOWN:
-        drawing = True
-        roi_points = [(x, y)]
-    elif event == cv2.EVENT_MOUSEMOVE:
-        if drawing:
-            roi_points.append((x, y))
-    elif event == cv2.EVENT_LBUTTONUP:
-        drawing = False
-        roi_points.append((x, y))
-
-# Function to perform image segmentation using the watershed algorithm
-def segment_image(input_image):
-    # Convert the input image (NumPy array) to a format that OpenCV can use
-    image = input_image.astype(np.uint8)
-    
-    # Create a copy of the image for manual segmentation
-    manual_segmentation_image = image.copy()
-    
-    # Create a mask for manual segmentation
-    mask = np.zeros(image.shape[:2], dtype=np.uint8)
-    
-    # Create a window for manual segmentation
-    cv2.namedWindow("Manual Segmentation")
-    cv2.setMouseCallback("Manual Segmentation", mouse_event)
-    
-    while True:
-        for point in roi_points:
-            cv2.circle(manual_segmentation_image, point, 5, (0, 0, 255), -1)
-        
-        cv2.imshow("Manual Segmentation", manual_segmentation_image)
-        key = cv2.waitKey(1) & 0xFF
-        
-        # Press 's' to perform segmentation
-        if key == ord("s"):
-            break
-        # Press 'r' to reset manual segmentation
-        elif key == ord("r"):
-            manual_segmentation_image = image.copy()
-            roi_points = []
-    
-    # Close the manual segmentation window
-    cv2.destroyWindow("Manual Segmentation")
-    
-    # Convert the image to grayscale
-    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+# Define a function for marker-based segmentation using the OpenCV watershed algorithm
+def watershed_segmentation(input_image, markers_image):
+    # Load the input image
+    image = cv2.imdecode(input_image, cv2.IMREAD_COLOR)
     
-    # Create a mask based on the manually segmented ROIs
-    if roi_points:
-        roi_points = np.array(roi_points, np.int32)
-        cv2.fillPoly(mask, [roi_points], 255)
+    # Load the marker image (binary image with markers)
+    markers = cv2.imdecode(markers_image, cv2.IMREAD_GRAYSCALE)
     
-    # Apply the watershed algorithm to segment the image
-    cv2.watershed(image, mask)
+    # Apply the watershed algorithm
+    image_copy = image.copy()
+    cv2.watershed(image_copy, markers)
     
-    # Create a segmented output image
-    output_image = image.copy()
-    output_image[mask == -1] = [0, 0, 255]  # Mark watershed boundaries in red
+    # Apply color mapping to the segmented regions
+    segmented_image = cv2.applyColorMap(markers, cv2.COLORMAP_JET)
     
-    return output_image
+    return segmented_image
 
-# Create a Gradio interface
-iface = gr.Interface(
-    fn=segment_image,
-    inputs="image",
-    outputs="image",
-    title="Manual Image Segmentation using Watershed Algorithm",
-    description="Upload an image and perform manual image segmentation by drawing regions of interest (ROIs) before segmenting. Press 's' to segment and 'r' to reset ROIs.",
-)
+# Define Gradio interfaces for the input and output
+input_image = gr.inputs.Image()
+markers_image = gr.inputs.Image()
+output_image = gr.outputs.Image()
 
-# Launch the Gradio app
-iface.launch()
+# Create a Gradio app
+gr.Interface(
+    fn=watershed_segmentation,
+    inputs=[input_image, markers_image],
+    outputs=output_image,
+    title="Marker-Based Segmentation with Watershed Algorithm",
+    description="Upload an image and a marker image to perform marker-based segmentation.",
+).launch()