Upload 3 files

app.py

@@ -1,65 +1,60 @@
 import os
 import gradio as gr
-import numpy as np
 from PIL import Image
+from model import kmeans, mean_shift
 
-#Fonction du modèle provisoire
-def segment_image(image):
-    overlay_image = image.copy()
-    return overlay_image
+with gr.Blocks() as demo:
+    gr.HTML(
+    """<style>
+    @import url('https://fonts.googleapis.com/css2?family=Poppins:wght@600&family=Roboto:wght@400&display=swap');
+
+    body { background-color: #f0f0f0; font-family: 'Roboto', sans-serif; }
+    h1 { font-family: 'Poppins', sans-serif; color: #4CAF50; text-align: center; }
+    h3 { font-family: 'Roboto', sans-serif; color: #333; text-align: center; }
+    .segment-btn { 
+        background-color: #4CAF50; 
+        border-radius: 8px; 
+        padding: 10px 20px; 
+        color: white; 
+        transition: background-color 0.3s ease, transform 0.2s ease;
+    }
+    .segment-btn:hover {
+        background-color: #45a049; 
+        transform: scale(1.05);
+    }
+    .logo { 
+        display: block; 
+        margin: 0 auto; 
+        width: 150px; 
+        height: auto; 
+        padding: 20px 0;
+    }
+    </style>"""
+    )
+    #logo_path = os.path.join(os.getcwd(), 'assets/img/fashionlook1.png')
+    gr.Markdown("""
+        <div style="text-align: center">
+            <img src='./fashionlook1.png' class='logo' alt='Logo' style='max-width: 300px'>
+        </div>
+        """)
+    gr.Markdown("<h1>FashionLook - Segment Clothes</h1>")
+    gr.Markdown("<h3 style='text-align: center;'>" "Upload an image and let our model detect and segment clothes such as shirts, pants, skirts...""</h3>")
+    with gr.Row():
+        with gr.Column(scale=1):
+            gr.Markdown("<h5>Upload your image</h5>")
+            image_input = gr.Image(type='pil', label="Upload Image")
+            with gr.Row():
+                segment_button = gr.Button("Run Segmentation", elem_id="segment-btn")
+        with gr.Column(scale=1):
+            gr.Markdown("<h5>Segmented Image with Overlay</h5>")
+            segmented_image_output = gr.Image(type="pil", label="Segmented Image", interactive=False)
 
-def create_interface():
-    with gr.Blocks() as interface:
-        gr.HTML(
-        """<style>
-        @import url('https://fonts.googleapis.com/css2?family=Poppins:wght@600&family=Roboto:wght@400&display=swap');
-        body { background-color: #2c2c2c; font-family: 'Roboto', sans-serif; }
-        h1 { font-family: 'Poppins', sans-serif; color: #ff9800; text-align: center; font-size: 2.5em; margin-bottom: 20px; }
-        h3 { font-family: 'Roboto', sans-serif; color: #dddddd; text-align: center; margin-bottom: 10px; }
-        h5 { font-family: 'Roboto', sans-serif; color: #dddddd; text-align: center; margin-bottom: 10px; }
-        .segment-btn { 
-            background-color: #ff9800; 
-            border-radius: 8px; 
-            padding: 10px 20px; 
-            color: white; 
-            transition: background-color 0.3s ease, transform 0.2s ease;
-        }
-        .segment-btn:hover {
-            background-color: #e67e00; 
-            transform: scale(1.05);
-        }
-        .logo { 
-            display: block; 
-            margin: 0 auto; 
-            width: 150px; 
-            height: auto; 
-            padding: 20px 0;
-        }
-        </style>"""
-        )
-        #logo_path = os.path.join(os.getcwd(), 'fashionlookl1_2.png')
-        gr.Markdown(f"<img src='fashionlookl1_2.png' class='logo' alt='Logo'>")
-        gr.Markdown("<h1>FashionLook - Segment Clothes</h1>")
-        gr.Markdown("<h3 style='text-align: center;'>" "Upload an image and let our model detect and segment clothes such as shirts, pants, skirts...""</h3>")
-        with gr.Row():
-            with gr.Column(scale=1):
-                gr.Markdown("<h5>Upload your image</h5>")
-                image_input = gr.Image(type='pil', label="Upload Image")
-                with gr.Row():
-                    segment_button = gr.Button("Run Segmentation", elem_id="segment-btn")
-            with gr.Column(scale=1):
-                gr.Markdown("<h5>Segmented Image with Overlay</h5>")
-                segmented_image_output = gr.Image(type="pil", label="Segmented Image", interactive=False)
-        
         # Actions liées aux inputs/outputs
         segment_button.click(
-            fn=segment_image, 
+            fn=kmeans, 
             inputs=[image_input], 
             outputs=[segmented_image_output]
         )
-    
-    return interface
 
 # Lancer l'interface
-interface = create_interface()
-interface.launch()
+demo.launch(share=False)

model.py

@@ -0,0 +1,32 @@
+import numpy as np
+import cv2
+
+def kmeans(image, k=5, alpha=0.5):
+    image = np.array(image)
+    pixel_vals = image.reshape((-1,3))
+    pixel_vals = np.float32(pixel_vals)
+
+    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 100, 0.85)
+    retval, labels, centers = cv2.kmeans(pixel_vals, k, None, criteria, 10, cv2.KMEANS_RANDOM_CENTERS)
+
+    centers = np.uint8(centers)
+    #  Attribuer des couleurs données à nos clusters
+    cluster_colors = np.random.randint(0, 255, size=(k, 3), dtype=np.uint8)
+    segmented_data = cluster_colors[labels.flatten()]
+    segmented_image = segmented_data.reshape((image.shape))
+
+    return segmented_image
+
+def mean_shift(image, spatial_radius=5, color_radius=60, max_iter=4):
+    image = np.array(image)
+    mean_shift_result = cv2.pyrMeanShiftFiltering(image, sp=spatial_radius, sr=color_radius, maxLevel=max_iter)
+
+    flat_image = mean_shift_result.reshape((-1, 3))
+
+    unique_colors, labels = np.unique(flat_image, axis=0, return_inverse=True)
+
+    cluster_colors = np.random.randint(0, 255, size=(len(unique_colors), 3), dtype=np.uint8)
+    segmented_image = cluster_colors[labels].reshape(image.shape)
+
+    return segmented_image
+